article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
microwave rb vapor - cell atomic clocks @xcite , based on optical - microwave double resonance , are today ubiquitous timing devices used in numerous fields of industry including instrumentation , telecommunications or satellite - based navigation systems .
their success is explained by their ability to demonstrate excellent short - term fractional frequency stability at the level of 10@xmath1 , combined with a small size , weight , power consumption and a relatively modest cost . over the last decade ,
the demonstration of advanced atom interrogation techniques ( including for instance pulsed - optical - pumping ( pop ) ) using narrow - linewidth semiconductor lasers has conducted to the development in laboratory of new - generation vapor cell clocks @xcite .
these clocks have succeeded to achieve a 100 times improvement in frequency stability compared to existing commercial vapor cell clocks . in this domain ,
clocks based on a different phenomenon , named coherent population trapping ( cpt ) , have proven to be promising alternative candidates . since its discovery in @xmath2 @xcite , coherent population trapping physics @xcite has motivated stimulating studies in various fields covering fundamental and applied physics such as slow - light experiments @xcite , high - resolution laser spectroscopy , magnetometers @xcite , laser cooling @xcite or atomic frequency standards .
basically , cpt occurs by connecting two long - lived ground state hyperfine levels of an atomic specie to a common excited state by simultaneous action of two resonant optical fields . at null raman detuning , i.e. when the frequency difference between both optical fields matches perfectly the atomic ground - state hyperfine frequency , atoms are trapped through a destructive quantum interference process into a noninteracting coherent superposition of both ground states , so - called dark state , resulting in a clear decrease of the light absorption or equivalently in a net increase of the transmitted light .
the output resonance signal , whose line - width is ultimately limited by the cpt coherence lifetime , can then be used as a narrow frequency discriminator towards the development of an atomic frequency standard . in a cpt - based clock , unlike the traditional double - resonance rb clock @xcite , the microwave signal used to probe the hyperfine frequency is directly optically carried allowing to remove the microwave cavity and potentially to shrink significantly the clock dimensions .
the application of cpt to atomic clocks was firstly demonstrated in a sodium atomic beam @xcite . in @xmath3 , n. cyr
et al proposed a simple method to produce a microwave clock transition in a vapor cell with purely optical means by using a modulated diode laser @xcite , demonstrating its high - potential for compactness . in @xmath4 , a first remarkably compact atomic clock prototype was demonstrated in nist @xcite .
further integration was achieved later thanks to the proposal @xcite and development of micro - fabricated alkali vapor cells @xcite , leading to the demonstration of the first chip - scale atomic clock prototype ( csac ) @xcite and later to the first commercially - available csac @xcite .
nevertheless , this extreme miniaturization effort induces a typical fractional frequency stability limited at the level of @xmath5 , not compliant with dedicated domains requiring better stability performances . in that sense , in the frame of the european collaborative mclocks project @xcite
, significant efforts have been pursued to demonstrate compact high - performance cpt - based atomic clocks and to help to push this technology to industry . in standard cpt clocks ,
a major limitation to reach better frequency stability performances is the low contrast ( @xmath6 , the amplitude - to - background ratio ) of the detected cpt resonance .
this low contrast is explained by the fact that atoms interact with a circularly polarized bichromatic laser beam , leading most of the atomic population into extreme zeeman sub - levels of the ground state , so called `` end - states '' .
several optimized cpt pumping schemes , aiming to maximize the number of atoms participating to the clock transition , have been proposed in the literature to circumvent this issue ( @xcite , and references therein ) , but at the expense of increased complexity . in that sense , a novel constructive polarization modulation cpt @xcite pumping technique , named double - modulation ( dm ) scheme , was recently proposed .
it consists to apply a phase modulation between both optical components of the bichromatic laser synchronously with a polarization modulation .
the phase modulation is needed to ensure a common dark state to both polarizations , allowing to pump a maximum number of atoms into the desired magnetic - field insensitive clock state .
this elegant solution presents the main advantage compared to the push - pull optical pumping @xcite or the lin@xmath7lin technique @xcite to avoid any optical beam separation or superposition and is consequently well - adapted to provide a compact and robust linear architecture setup . in this article
, we demonstrate a high - performance cw - regime cpt clock based on the dm technique .
optimization of the short - term frequency stability is performed by careful characterization of the cpt resonance versus relevant experimental parameters .
a short - term frequency stability at the level of @xmath0 up to 100 s , comparable to best vapor cell frequency standards , is reported .
a detailed noise budget is given , highlighting a dominant contribution of the microwave power fluctuations .
section ii describes the experimental setup .
section iii reports the detailed cpt resonance spectroscopy versus experimental parameters .
section iv reports best short - term frequency stability results .
noise sources limiting the stability are carefully analysed . in section
v , we study the clock frequency shift versus each parameter and estimate the limitation of the clock mid - term frequency stability .
our setup is depicted in fig .
a dfb laser diode emits a monochromatic laser beam around , the wavelength of the cs @xmath8 line . with the help of a fiber electro - optic phase modulator ( eopm ) , modulated at with about microwave power , about @xmath9 of the carrier power
is transferred into both first - order sidebands used for cpt interaction .
the phase between both optical sidebands , so - called raman phase in the following , is further modulated through the driving microwave signal .
two acousto - optic modulators ( aoms ) are employed .
the first one , aom1 , is used for laser power stabilization .
the second one , aom2 , allows to compensate for the buffer - gas induced optical frequency shift ( @xmath10 ) in the cpt clock cell .
a double - modulated laser beam is obtained by combining the phase modulation with a synchronized polarization modulation performed thanks to a liquid crystal polarization rotator ( lcpr ) .
the laser beam is expanded to before the vapor cell .
the cylindrical cs vapor cell , diameter and long , is filled with of mixed buffer gases ( argon and nitrogen ) .
unless otherwise specified , the cell temperature is stabilized to about . a uniform magnetic field of is applied along the direction of the cell axis by means of a solenoid .
the ensemble is surrounded by two magnetic shields in order to remove the zeeman degeneracy .
half(quarter)-wave plate , bg buffer gas , pd photodiode.,scaledwidth=45.0% ] we first utilized a fabry - perot cavity to investigate the eopm sidebands power ratio versus the coupling microwave power ( @xmath11 ) , see fig .
[ sideband ] .
we choose @xmath11 around to maximize the power transfer efficiency into the first - order sidebands .
the sidebands spectrum is depicted in the inset of fig .
[ sideband ] . obtained by scanning the fp cavity length ,
notice the log scale of the y - axis .
, scaledwidth=45.0% ] since the laser intensity noise is known as being one of the main noise sources which limit the performances of a cpt clock @xcite , laser power needs to be carefully stabilized . for this purpose ,
a polarization beam splitter ( pbs ) reflects towards a photo - detector a part of the laser beam , the first - order diffracted by aom1 following the eopm .
the output voltage signal is compared to an ultra - stable voltage reference ( lt1021 ) .
the correction signal is applied on a voltage variable power attenuator set on the feeding rf power line of the aom1 with a servo bandwidth of about . the out - loop laser intensity noise ( rin )
is measured just after the first pbs with a photo - detector ( pd ) , which is not shown in fig .
[ fig1 ] . the spectrum of the resulting rin with and w / o locking is shown in fig . [ fig3 ] .
a improvement at @xmath12 ( lo modulation frequency for clock operation ) is obtained in the stabilized regime .
it is worth to note that the dfb laser diode we used , with a linewidth of about , is sensitive even to the lowest levels of back - reflections @xcite , e.g. , the coated collimated lens may introduce some intensity and frequency noise at the regime of .
finding the correct lens alignment to minimize the reflection induced noise while keeping a well - collimated laser beam was not an easy task . to reduce light feedback from the eopm fiber face
, we use a isolator before the eopm .
the fiber - coupled eopm induces additional intensity noise depicted in fig .
nevertheless , thanks to the laser power locking , we can reduce most of these noises by at least in the range of to .
line in the vacuum reference cell and in the clock cell recorded with the bichromatic laser .
the two absorptions from left to right correspond to the excited level @xmath13 and @xmath14 , respectively .
for the reference cell signal : laser power , beam diameter , cell temperature , aom frequency .
inset , the atomic levels involved in @xmath8 line of cesium . , scaledwidth=45.0% ] our laser frequency stabilization setup , similar to @xcite , is depicted in fig .
we observe in a vacuum cesium cell the two - color doppler - free spectrum depicted in fig .
the bi - chromatic beam , linearly polarized , is retro - reflected after crossing the cell with the orthogonal polarization .
only atoms of null axial velocity are resonant with both beams .
consequently , cpt states built by a beam are destroyed by the reversed beam , leading to a doppler - free enhancement of the absorption @xcite .
the laser frequency detuning @xmath15 in fig .
[ fig4 ] corresponds to the laser carrier frequency tuned to the center of both transitions @xmath16 and @xmath17 in d@xmath18 line of cesium , where @xmath19 is the hyperfine quantum number . for this record ,
the microwave frequency is , half the cs ground state splitting , and the dfb laser frequency is scanned . the frequency noise with and w / o locking are presented in fig . [ fig5 ] .
the servo bandwidth is about and the noise is found to be reduced by about at @xmath20 = ( local oscillator modulation frequency in clock operation ) .
we studied the response time of the lcpr ( fpr-100 - 895 , meadowlark optics ) .
as illustrated in fig .
[ fig6 ] , the measured rise ( fall ) time is about and the polarization extinction ratio is about @xmath21 . in comparison ,
the electro - optic amplitude modulator ( eoam ) used as polarization modulator in our previous investigations @xcite showed a response time of ( limited by our high voltage amplifier ) and a polarization extinction ratio of @xmath22 . here , we replace it by a liquid crystal device because its low voltage and small size would be an ideal choice for a compact cpt clock , and we will show in the following that the longer switching time does not limit the contrast of the cpt signal .
the electronic system ( local oscillator and digital electronics for clock operation ) used in our experiment is depicted in fig .
[ microwave - chain ] .
the microwave source is based on the design described in @xcite .
the local oscillator ( lo ) is a module ( xm16 pascall ) integrating an ultra - low phase noise 100 mhz quartz oscillator frequency - multiplied without excess noise to .
the signal is synthesized by a few frequency multiplication , division and mixing stages .
the frequency modulation and tuning is yielded by a direct digital synthesizer ( dds ) referenced to the lo .
the clock operation @xcite is performed by a single field programmable gate array ( fpga ) which coordinates the operation of the dds , analog - to - digital converters ( adc ) and digital - to - analog converter ( dacs ) : \(1 ) the dds generates a signal with phase modulation ( modulation rate @xmath23 , depth @xmath24 ) and frequency modulation ( @xmath25 , depth @xmath26 ) .
\(2 ) the dac generates a square - wave signal to drive the lcpr with the same rate @xmath23 , synchronous to the phase modulation .
\(3 ) the adc is the front - end of the lock - in amplifier .
another dac , used to provide the feedback to the local oscillator frequency , is also implemented in the fpga .
the clock frequency is measured by comparing the lo signal with a signal delivered by a h maser of the laboratory in a symmetricom @xmath27a allan deviation test set .
the frequency stability of the maser is @xmath28 at integration time .
modulation frequency of the signal , @xmath23 polarization and phase modulation frequency , @xmath29 pumping time , @xmath30 detection window.,scaledwidth=45.0% ] as illustrated in fig . [ fig8 ] , the polarization and phase modulation share the same modulation function . after a pumping time @xmath29
to prepare the atoms into the cpt state , we detect the cpt signal with a window of length @xmath30 . in order to get an error signal to close the clock frequency loop ,
the microwave frequency is square - wave modulated with a frequency @xmath25 , and a depth @xmath26 . in our case , we choose @xmath31 hz , as a trade - off between a low frequency to have time to accumulate the atomic population into the clock states by the dm scheme and a high operating frequency to avoid low frequency noise in the lock - in amplification process and diminish the intermodulation effects . clock transition . working parameters : @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 , @xmath37 .
, scaledwidth=45.0% ] a typical experimental cpt signal , recorded with this time sequence , showing all the cpt transitions allowed between zeeman sub - levels of the cs ground state is reported in fig .
[ zeeman ] .
the raman detuning @xmath38 is the difference between the two first sideband spacing and the cs clock resonance .
the spectrum shows that the clock levels ( @xmath39 transition ) are the most populated and that the atomic population is symmetrically distributed around the @xmath40 sub - levels .
the distortion of neighbouring lines is explained by magnetic field inhomogeneities .
it can be shown that the clock short - term frequency stability limited by an amplitude noise scales as @xmath41 @xcite , with @xmath42 the full width at half maximum ( fwhm ) of the clock resonance , and @xmath6 the contrast of the resonance .
usually , the ratio of contrast @xmath6 to @xmath42 is adopted as a figure of merit , _
i.e. _ @xmath43 .
the best stability should be obtained by maximizing @xmath44 .
the stability of the clock is measured by the allan standard deviation @xmath45 , with @xmath46 the averaging time .
when the signal noise is white , with standard deviation @xmath47 , and for a square - wave frequency modulation , the stability limited by the signal - to - noise ratio is equal to @xcite @xmath48 with @xmath49 the clock frequency , and @xmath50 the slope of the frequency discriminator . in cpt clocks ,
one of the main sources of noise is the laser intensity noise , which leads to a signal noise proportional to the signal .
therefore it is more convenient to characterize the quality of the signal of a cpt atomic clock by a new figure of merit , @xmath51 , where @xmath50 is the slope of the error signal ( in v / hz ) at raman resonance ( @xmath52 ) , and @xmath53 is the detected signal value ( in v ) at the interrogating frequency ( the clock resonance frequency plus the modulation depth @xmath26 ) , see fig .
[ cptsignal ] .
note that an estimation of the discriminator slope is also included in @xmath44 , since the contrast is the signal amplitude @xmath54 divided by the background @xmath55 .
@xmath44 then equals @xmath56 , @xmath57 is a rough approximation of the slope @xmath50 , and @xmath55 an approximation of the working signal @xmath53 . in our experimental conditions @xmath58 . , @xmath59 , @xmath60 , @xmath35 , @xmath36 , @xmath37.,scaledwidth=45.0% ] we investigated the effect of relevant parameters on both figures of merit to optimize the clock performances .
in order to allow a comparison despite different conditions , the error signals are generated with the same unit gain .
since the resonance linewidth is also subject to change , it is necessary to optimize the 4.6 ghz modulation depth @xmath26 to maximize @xmath61 . here for simplicity , we first recorded the cpt signal , then we can numerically compute optimized values of @xmath26 and @xmath61 . in the following ,
we investigate the dependence of @xmath44 and @xmath61 on several parameters including the cell temperature ( @xmath62 ) , the laser power ( @xmath63 ) , the microwave power ( @xmath11 ) , the detection window duration ( @xmath30 ) , the detection start time ( @xmath29 ) , and the polarization and phase modulation frequency ( @xmath23 ) . , @xmath44 , contrast @xmath6 and width of the clock transition as function of cell temperature @xmath62 .
all other working parameters is the same as fig .
10.,scaledwidth=45.0% ] from the figures of merit shown in fig .
[ tcell ] , the optimized cell temperature is around @xmath64 for @xmath35 .
the narrower linewidth observed at higher @xmath62 , already observed by godone _
@xcite , can be explained by the propagation effect : the higher the cell temperature , the stronger the light absorption by more atoms , and less light intensity is seen by the atoms at the end side of the vapour cell .
this leads to a reduction of the power broadening and a narrower signal , measured by the transmitted light amplitude .
the optimum temperature depends on the laser power as depicted in fig .
[ tcell-2 ] .
nevertheless , the overall maximum of @xmath61 is reached with @xmath35 at @xmath37 . as a function of cell temperature @xmath62 for various laser powers .
all other working parameters is the same as fig .
, scaledwidth=45.0% ] the figures of merit @xmath61 , @xmath44 , the contrast @xmath6 and the width are plotted as a function of the laser power in fig .
[ pl ] for @xmath37 .
the laser powers maximizing @xmath61 and @xmath44 are @xmath35 and @xmath65 , respectively .
the allan deviation reaches a better value at @xmath35 , which justifies our choice of @xmath61 as figure of merit . and in the following parameters investigation
, we only show the @xmath61 as figure of merit for clarity . , @xmath44 , @xmath6 and width of the clock transition as a function of laser power @xmath63 with @xmath37 .
all other working parameters is the same as fig .
10.,scaledwidth=45.0% ] , @xmath6 and width of the clock transition as function of microwave power @xmath11 with @xmath37 .
all other working parameters is the same as fig .
, scaledwidth=45.0% ] @xmath61 , @xmath6 and width versus the microwave power are shown in fig .
the behaviour of @xmath61 is basically in agreement with the fractional power of first ( @xmath66 ) sidebands of fig .
[ sideband ] .
the optimized microwave power is around @xmath67 . as we can see on fig .
[ tw ] , a short detection window @xmath30 would generate a higher contrast signal and higher figures of merit .
however , we found that a longer time , e.g. , @xmath59 , results in a better allan deviation at one - second averaging time .
it is due to the conflict between the higher signal slope ( @xmath68 ) and the increased number of detected samples which help to reduce the noise , see eq.([sigma_yp_i-3 ] ) . , @xmath6 and width of the clock transition as a function of @xmath30 . with @xmath69 and all other working parameters
is the same as fig .
10.,scaledwidth=45.0% ] the same parameters are plotted versus the pumping time @xmath29 in fig .
the figures of merit and @xmath6 firstly increase and then decrease at @xmath70 , because when the detection window @xmath71 , the signals of two successive polarizations are included .
the dynamic behaviour of the atomic system induces the decrease of the cpt amplitude ( see fig .
10 in @xcite ) .
after @xmath32 , @xmath61 increases again and reach a maximum .
thus we can say that , in a certain range , a longer @xmath29 will lead to a greater atomic population pumped into the clock states as depicted in fig .
[ td ] , yielding higher figures of merit .
the behaviour of the linewidth versus the pumping time is not explained to date .
nevertheless , note that here the steady - state is not reached and the width behaviour results certainly from a transient effect . ,
@xmath6 and width of the clock transition versus @xmath29 .
all other working parameters is the same as fig .
10.,scaledwidth=45.0% ] figure [ fm ] shows @xmath61 , @xmath6 , and width versus the polarization ( and phase ) modulation frequency @xmath23 the maxima of @xmath61 is reached at low frequency @xmath23 . in one hand
, this is an encouraging result to demonstrate the suitability of the lcpr polarization modulator in this experiment . in an other hand
, the higher @xmath25 rate would be better for a clock operation with lock - in method to modulate and demodulate the error signal , to avoid the low frequency noises such as @xmath72 noise .
therefore , we chose @xmath73 and @xmath60 .
we have noticed that the behavior of @xmath6 is not exactly the same than the one observed in our previous work @xcite with a fast eoam , where the signal amplitude was maximized at higher frequencies .
this can be explained by the slower response time of the polarization modulator and the lower laser intensity used .
the linewidth reaches a minimum around @xmath74 .
this behaviour will be investigated in the future . ,
@xmath6 and width of the clock transition as function of @xmath23 .
all other working parameters is the same as fig .
the high contrast and narrow line - width cpt signal obtained with the optimized values of the parameters is presented in fig .
[ cptsignal ] , with the related error signal .
the allan standard deviation of the free - running lo and of the clock frequency , measured against the h maser , are shown in fig .
[ allan ] .
the former is in correct agreement with its measured phase noise . in the offset frequency region , the phase noise spectrum of the free - running lo signal
is given in by @xmath75 with @xmath76 = @xmath7747 , signature of a flicker frequency noise @xcite .
this phase noise yields an expected allan deviation given by @xmath78 ( ) @xmath10 @xmath79 @xmath10 1.2@xmath80 @xcite , close to the measured value of @xmath81 at
. the measured stability of the cpt clock is @xmath0 up to averaging time for our best record .
this value is close to the best cpt clocks @xcite , demonstrating that a high - performance cpt clock can be built with the dm - cpt scheme .
a typical record for longer averaging times is also shown in fig .
[ allan ] . for averaging times @xmath46 longer than , the allan deviation increases like @xmath82 , signature of a random walk frequency noise .
( @xmath83 ) , respectively . , scaledwidth=45.0% ] we have investigated the main noise sources that limit the short - term stability . for a first estimation , we consider only white noise sources , and for the sake of simplicity we assume that the different contributions can independently add , so that the total allan variance can be computed as @xmath84 with @xmath85 the contribution due to the phase noise of the local oscillator , and @xmath86 the allan variance of the clock frequency induced by the fluctuations of the parameter @xmath87 .
when @xmath87 modifies the clock frequency during the whole interrogation cycle , @xmath86 can be written @xmath88 with @xmath89 the variance of @xmath87 measured in 1 hz bandwidth at the modulation frequency @xmath25 , @xmath90 is the clock frequency sensitivity to a fluctuation of @xmath87 . here , the detection signal is sampled during a time window @xmath30 with a sampling rate @xmath91 , where @xmath92 is a cycle time . in this case eq.([sigma_yp_i ] ) becomes @xmath93 with @xmath94 the variance of @xmath87 sampled during @xmath30 ; @xmath95 with @xmath96 the value of the power spectral density ( psd ) of @xmath87 at the fourier frequency @xmath25 ( assuming a white frequency noise around @xmath25 ) . when @xmath87 induces an amplitude fluctuation with a sensitivity @xmath97 , eq.([sigma_yp_i-2 ] ) becomes @xmath98 with @xmath50 the slope of the frequency discriminator in v / hz . we review below the contributions of the different sources of noise . detector noise : the square root of the power spectral density ( psd ) @xmath99 of the signal fluctuations measured in the dark is shown in fig . [ snr ]
it is @xmath100 nv@xmath101 in @xmath102 bandwidth at the fourier frequency . according to eq.([sigma_yp_i-3 ] )
the contribution of the detector noise to the allan deviation at one second is @xmath103 .
shot - noise : with the transimpedance gain @xmath104 and the detector current @xmath105 , eq.([sigma_yp_i-3 ] ) becomes @xmath106 with @xmath107 the electron charge .
the contribution to the allan deviation at one second is .
laser fm - am noise : it is the amplitude noise induced by the laser carrier frequency noise .
the slope of the signal @xmath53 with respect to the laser frequency @xmath108 is @xmath109 at optical resonance . according to eq.([sigma_yp_i-3 ] ) with data of laser - frequency - noise psd of fig .
[ fig5 ] at 125 hz , we get a allan deviation of at one second .
laser am - am noise : it is the amplitude noise induced by the laser intensity noise .
the measured signal sensitivity to the laser power is @xmath110 at @xmath111 , combined with the laser intensity psd of fig .
[ fig3 ] it leads to the amplitude noise @xmath112 nv@xmath101 , and an allan deviation of at one second .
lo phase noise : the phase noise of the local oscillator degrades the short - term frequency stability via the intermodulation effect @xcite . it can be estimated by : @xmath113 our microwave source is based on @xcite which shows an ultra - low phase noise @xmath114 at @xmath115 fourier frequency .
this yields a contribution to the allan deviation of at one second .
microwave power noise : fluctuations of microwave power lead to a laser intensity noise , which is already taken into account in the rin measurement .
we show in the next section that they also lead to a frequency shift ( see fig .
[ f0-puw ] ) .
the allan deviation of the microwave power at @xmath116 is @xmath117 , see inset of fig .
[ f0-puw ] . with a measured slope of @xmath118 hz / dbm
, we get a fractional - frequency allan deviation of , which is the largest contribution to the stability at .
note that in our set - up the microwave power is not stabilized . + the other noise sources considered have much lower contributions , they are the laser frequency - shift effect , _
i.e. _ am - fm and fm - fm contributions , the cell temperature and the magnetic field . table [ 1s ] resumes the short - term stability noise budget .
.[tab : table1 ] noise contributions to the stability at .
[ cols="^,^,^",options="header " , ] [ 1000s ]
we have implemented a compact vapor cell atomic clock based on the dm cpt technique . a detailed characterization of the cpt resonance versus several experimental parameters was performed .
a clock frequency stability of @xmath0 up to averaging time was demonstrated .
for longer averaging times , the allan deviation scales as @xmath82 , signature of a random walk frequency noise .
it has been highlighted that the main limitation to the clock short and mid - term frequency stability is the fluctuations of the microwave power feeding the eopm .
improvements could be achieved by implementing a microwave power stabilization .
another or complementary solution could be to choose a finely tuned laser power value minimizing the microwave power sensitivity .
this adjustment could be at the expense of the signal reduction and a trade - off has to be found .
nevertheless , the recorded short - term stability is already at the level of best cpt clocks @xcite and close to state - of - the art rb vapor cell frequency standards .
these preliminary results show the possibility to a high - performance and compact cpt clock based on the dm - cpt technique .
we thank moustafa abdel hafiz ( femto - st ) , david holleville and luca lorini ( lne - syrte ) for helpful discussions .
we are also pleased to acknowledge charles philippe and ouali acef for supplying the thermal insulation material , michel abgrall for instrument symmetricom 5125a lending , david horville for laboratory arrangement , jos pinto fernandes , michel lours for electronic assistance , pierre bonnay and annie grard for manufacturing cs cells .
p. y. is supported by the facilities for innovation , research , services , training in time & frequency ( labex first - tf ) .
this work is supported in part by anr and dga(isimac project anr-11-astr-0004 ) .
this work has been funded by the emrp program ( ind55 mclocks ) .
the emrp is jointly funded by the emrp participating countries within euramet and the european union .
t. bandi , c. affolderbach , c. stefanucci , f. merli , a. k. skrivervik and g. mileti , coninuous - wave double - resonance rubidium standard with @xmath119 stability , ieee ultrason .
. contr . * 61 * , 11 , 17691778 ( 2014 ) .
s. kang , m. gharavipour , c. affolderbach , f. gruet , and g. mileti , demonstration of a high - performance pulsed optically pumped rb clock based on a compact magnetron - type microwave cavity , j. appl
. phys . * 117 * , 104510 ( 2015 ) .
j. e. thomas , s. ezekiel , c. c. leiby , r. h. picard , and c. r. willis , ultrahigh - resolution spectroscopy and frequency standards in the microwave and far - infrared regions using optical lasers , opt . lett . * 6 * , 298 - 300 ( 1981 ) .
j. e. thomas , p. r.hemmer , s. ezekiel , c. c. leiby , , r. h. picard , and c. r. willis , observation of ramsey fringes using a stimulated , resonance raman transition in a sodium atomic beam , phys .
lett . * 48 * , 867 - 870 ( 1982 ) .
j. kitching , n. vukicevic , l. hollberg , s. knappe , r.wynands , and w. weidemann , a microwave frequency reference based on vcsel - driven dark line resonances in cs vapor , ieee trans .
. measur . * 49 * , 1313 - 1317 ( 2000 ) .
v. shah and j. kitching , advances in coherent popualtion trapping for atomic clocks , in advances in atomic , molecular , and optical physics , edited by e. arimondo , p. r. berman , and c. c. lin , vol
. 59 ( elsevier , amsterdam , 2010 ) .
peter yun , doctor thesis , exploring new approaches to coherent population trapping atomic frequency standards , wuhan institute of physics and mathematics chinese academy of sciences , wuhan , china , 2012 .
x. liu , j .-
mrolla , s. gurandel , c. gorecki , e. de clercq , and r. boudot , coherent population trapping resonances in buffer - gas - filled cs - vapor cells with push - pull optical pumping , phys .
a * 87 * , 013416 ( 2013 ) .
t. zanon , s. gurandel , e. de clercq , d. holleville , n. dimarcq and a. clairon , high contrast ramsey fringes with coherent - population - trapping pulses in a double lambda atomic system , phys .
* 94 , 193002 ( 2005 ) .
* r.schmeissner , n. von bande , a.douahi , o.parillaud , m.garcia , m.krakowski , m.baldy , the optical feedback spatial phase driving perturbations of dfb laser diodes in an optical clock , proceedings of the 2016 european frequency and time forum , york ( 2016 ) , available at http://www.eftf.org/previousmeetings.php .
b. franois , c. e. calosso , m. abdel hafiz , s. micalizio , and r. boudot , simple - design ultra - low phase noise microwave frequency synthesizers for high - performing cs and rb vapor - cell atomic clocks , rev .
instrum . * 86 * , 094707 ( 2015 ) . c. e. calosso , s. micalizio , a. godone , e. k. bertacco , f. levi .
electronics for the pulsed rubidium clock : design and characterization .
ieee trans .
control * 54 * , 1731 - 1740 ( 2007 ) .
p. yun , s. mejri , f. tricot , m. abdel hafiz , r. boudot , e. de clercq , s. gurandel , double - modulation cpt cesium compact clock , 8th symposium on frequency standards and metrology 2015 , j. of physics : conference series * 723 * , 012012 ( 2016 ) .
m. zhu and l.s .
cutler , theoretical and experimental study of light shift in a cpt - based rb vapor cell frequency standard , in proceedings of the 32nd precise time and time interval systems and applications meeting , p. 311
, ed . by l.a .
breakiron ( us naval observatory , washington , dc , 2000 ) . c. affolderbach ,
c. andreeva , s. cartaleva , t. karaulanov , g. mileti , and d. slavov , light - shift suppression in laser optically pumped vapour - cell atomic frequency standards , appl .
b * 80 * , 841 - 8 ( 2005 ) . v. shah , v. gerginov , p. d. d. schwindt , s. knappe , l. hollberg and j. kitching , continuous light- shift correction in modulated coherent population trapping clocks , appl . phys . lett . * 89 , 151124 ( 2006 ) . * | we demonstrate a vapor cell atomic clock prototype based on continuous - wave ( cw ) interrogation and double - modulation coherent population trapping ( dm - cpt ) technique .
the dm - cpt technique uses a synchronous modulation of polarization and relative phase of a bi - chromatic laser beam in order to increase the number of atoms trapped in a dark state , _
i.e. _ a non - absorbing state .
the narrow resonance , observed in transmission of a cs vapor cell , is used as a narrow frequency discriminator in an atomic clock . a detailed characterization of the cpt resonance versus numerous parameters is reported .
a short - term frequency stability of @xmath0 up to averaging time is measured .
these performances are more than one order of magnitude better than industrial rb clocks and comparable to those of best laboratory - prototype vapor cell clocks .
the noise budget analysis shows that the short and mid - term frequency stability is mainly limited by the power fluctuations of the microwave used to generate the bi - chromatic laser .
these preliminary results demonstrate that the dm - cpt technique is well - suited for the development of a high - performance atomic clock , with potential compact and robust setup due to its linear architecture .
this clock could find future applications in industry , telecommunications , instrumentation or global navigation satellite systems . |
from a mathematical viewpoint , anti - symmetrical tensors are natural objects living in differential manifolds .
they play an important role in the construction of the manifold s volume and therefore its orientation .
the dimension of the manifold defines the sort of possible anti - symmetrical tensors and this is used to define a specific space only of forms @xcite . besides this
, they are related to the linking number of higher dimensional knots @xcite . from the physical viewpoint , they are of great interest because they may have the status of fields describing particles other than the usual ones . as an example we can cite the space - time torsion @xcite and the axion field @xcite that have separated descriptions by the two - form .
besides this , string theory shows the naturalness of higher rank tensor fields in its spectrum @xcite .
other applications of these kind of fields have been made showing its relation with the ads / cft conjecture @xcite .
topological defects soliton - like are studied with increasing interest in physics , not only in condensed matter , as in particle physics and cosmology . in brane models , they are used as mechanism of fields localization , avoiding the appearance of the troublesome infinities .
several kinds of these defects in brane scenarios are considered in the literature @xcite . in these papers ,
the authors consider brane world models where the brane is supported by a soliton solution to the baby skyrme model or by topological defects available in some models . as an example , in a recent paper , a model is considered for coupling fermions to brane and/or antibrane modeled by a kink antikink system @xcite .
the subject of this article is to study aspects of the three - form field in a scenario of extra dimensions .
specifically we will study the issue of localization using solitons kink - like and the possibility of resonant kk modes in a randall - sundrum - like model ( a model with thick branes ) @xcite .
the core idea of extra dimensional models is to consider the four - dimensional universe as a hyper - surface embedded in a multidimensional manifold .
the appeal of such models is the determination of scenarios where membranes have the best chances to mimic the standard model s characteristics . in particular
, the standard model presents interesting topics to study such as the hierarchy problem , and the cosmological constant problem that can be treated by the above - mentioned scenarios .
for example , the randall - sundrum model @xcite provides a possible solution to the hierarchy problem and show how gravity is trapped to a membrane . in mathematical terms
, the presence of one more extra dimension ( @xmath8 ) provides the existence of many antisymmetric fields , namely the two , three , four and five forms .
however , the only relevant ones for our brane are the two and the three form .
this is due to the fact that when the number of dimensions increase , also increases the number of gauge freedom .
this can be used to cancel the dynamics of the field in the visible brane .
the mass spectrum of the two and three form have been studied , for example , in refs . @xcite and @xcite .
posteriorly , the coupling between the two and three forms with the dilaton was studied , in different contexts , in @xcite .
the study of this kind of coupling , inspired in string theory , is important in order to produce a process that , in principle , could be seen in lhc .
this is a drell - yang process in which a pair quark - antiquark can give rise to a three(two)-form field , mediated by a dilaton . in the present case ,
the coupling is needed for a different reason .
as we will see in this work , the three form field without a dilaton do not has its massless mode localized .
when we introduce a dilaton , this mode can be localized and therefore this is a more interesting case to be considered .
the localization of fields in a framework that consider the brane as a kink has been studied for example in @xcite .
this work is organized as follows .
the second section is devoted to the study of the membrane as a kink .
a solution of einstein equation is found for two scalar fields , one of them describing the kink solution and the other representing the dilaton . in the third section
we analyze the localization of the three form field with and without the dilaton , and observe that the localization is only possible in the framework with a dilaton coupling under a specific condition .
in the next section we study the gravitational background when a deformation procedure is carried out in our original framework .
we show that the massless mode of the three form is also localized only if we regard the dilaton coupling with the same condition cited above . in the fifth section
we make some considerations about the massive modes and the possibility of resonances . finally , in the last section , we discuss the conclusions and perspectives .
we start our analysis by studying the space - time background .
we go right to the point and give an explicit example through the localization of a zero mode of the three - form mater gauge field in a four - dimensional thick membrane embedded in five dimensions ( a randall - sundrum like scenario ) .
it is well known that vector gauge fields in these kind of scenarios are not localizable : in four dimensions the gauge vector field theory is conformal and all information coming from warp factors drops out necessarily rendering a non - normalizable four dimensional effective action . however , in the work of kehagias and tamvakis @xcite , it is shown that the coupling between the dilaton and the vector gauge field produces localization of the later . in analogy with the work of kehagias and
tamvakis we introduce here the coupling between the dilaton and the three - form field .
as we comment at the end of the next section , this coupling is also necessary to localize the three form field . before analyzing this coupling ,
it is necessary to obtain a solution of the equations of motion for the gravitational field in the background of the dilaton and the membrane . for such
, we introduce the following action @xcite : @xmath9.\ ] ] note again that we are working with a model containing two real scalar fields .
the field @xmath10 plays the role of to generate the membrane of the model while the field @xmath11 represents the dilaton .
the potential function now depends on both scalar fields .
it is assumed the following ansatz for the space - time metric : @xmath12 the equations of motion are given by @xmath13 @xmath14 @xmath15 and @xmath16 in order to solve that system , we use the so - called super - potential function @xmath17 , defined by @xmath18 , following the approach of kehagias and tamvakis @xcite .
the particular solution regarded follows from choosing the potential @xmath19 and super - potential @xmath20 as @xmath21 and @xmath22 in this way it is easy to obtain differential equations of first order whose solutions are solutions of the equations of motion above , namely @xmath23 @xmath24 and @xmath25 the solutions for these set of equations are the following : @xmath26 @xmath27 and @xmath28 following the argumentation in ref .
@xcite , it is possible to see that , by the linearization of the geometry described in this section , this model supports a massless zero mode of the gravitational field localized on the membrane , even in the dilaton background .
now the dilaton contribution makes the space - time singular .
however this kind of singularity is very common in d - brane solutions in string theory ( the dilaton solution is divergent ) .
the ricci scalar for this new geometry is now given by @xmath29 where the dilaton has an important contribution .
what is interesting here is that this singularity disappears if we lift the metric solution to @xmath30 . in this case
the dilaton represents the radius of the sixth dimension @xcite .
in this section we study antisymmetric tensor fields in the gravitational background with the dilaton field . we must look for localization of these fields in this framework . as the number of antisymmetric tensor fields increase with dimensions , these fields should be considered .
in fact these fields have been taken in account in the literature .
the fact is that , when the number of dimension increases , the number of gauge freedom also increases , and this can be used to cancel the degrees of freedom in the visible brane . therefore the only antisymmetric tensors relevant to the visible brane are that of rank two and three@xcite .
we must focus here in the second case .
we must specifically study the string inspired coupling of the three form field to the dilaton . as commented before this will give rise to the localization of the field . as commented in the introduction
, this kind of coupling can give an interaction that in principle could be seen at lhc . despite this , we focus here on the localization properties of this framework .
we have for the action @xmath31\ ] ] and @xmath32}$ ] is the field strength for the three form @xmath33 .
we can use gauge freedom to fix @xmath34 and we are left with the following terms @xmath35}\ ] ] @xmath36.\ ] ] using the above facts we obtain for the action @xmath37\},\end{aligned}\ ] ] with the equations of motion @xmath38=0.\ ] ] we can now separate the variables with the following ansatz @xmath39 where @xmath40 .
we now write @xmath41 , where @xmath42 stands for the four dimensional field strength and we get for the em @xmath43 a solution is found in the case @xmath44 , where we find that @xmath45 solves the above equation . in this case , the effective action can be found easily to give us @xmath46.\end{aligned}\ ] ] using now our solution for @xmath47 , @xmath48 and @xmath49 we have that , when @xmath50 , the integration in the extra dimension is finite .
therefore the massless mode of the three form field is possibly localized in a framework with a dilaton coupling .
it is worthwhile to analyze the case without the dilaton coupling . in this case
the metric is given by @xmath51 the gravitational solution has been found in @xcite and the warp factor is the same as in the case with dilaton coupling . therefore following the same steps as before , we can find the equation of motion @xmath52 again a direct solution is found in the case @xmath44 , and @xmath45 solves the above equation . in this particular case , the effective action can be found easily to give us @xmath53.\end{aligned}\ ] ] using now our solution for @xmath47 , we have that the integration in the extra dimension is not finite .
therefore we see that the dilaton coupling is really essential to the localization of the zero mode .
in this section we analyze a special class of solutions by defining a deformation of the @xmath54 potential @xcite .
it is possible to solve the equations of motion by the super - potential method .
this formalism was initially introduced in studies about non - super - symmetric domain walls in various dimensions by @xcite .
the first step in our analysis is to find the einstein s equations for the coupled system of the scalar - dilaton - gravity system that composes the background space - time .
we repeat here the same action already discussed in the sections above now with the super - potential @xmath55 where @xmath7 is an odd integer . by solving @xmath56
we arrive at @xmath57 and we see that for @xmath5 we get the usual kink solution .
for @xmath58 we can construct the so called two - kink solutions , describing internal structures inside the membrane @xcite .
the parameter @xmath7 introduced in the procedure controls characteristics such as thickness and matter energy density of the membrane @xcite .
we must note that the equations , and are left unchanged . from
the super - potential cited we get the first order equations @xmath59 @xmath60 and @xmath61 from the last equation we can find @xmath62 @xcite , @xmath63- \sum_{n=1}^{p-1}\frac1{2n}\tanh^{2n}\left(\frac{y}{p}\right)\biggr{\ } } \nonumber\end{aligned}\ ] ] and all the steps for obtaining the equations of motion and the effective action are basically identical . for the equation of motion we therefore obtain after a separation of variables @xmath64 and we can see that , again , we have a trivial solution for the case @xmath44 .
the effective action in the present case is easily found to be @xmath65.\end{aligned}\ ] ] using now our solution for @xmath66 , @xmath67 and @xmath68 we arrive in the same situation as before and have that , when @xmath69 and for finite @xmath7 , the integration in the extra dimension is finite .
therefore the massless mode of the three form field is possibly localized in a framework with a dilatonic deformed brane .
now we must analyze the possibility of localization for the massive modes in both , deformed and usual cases .
as the usual case can be obtained just by putting @xmath5 , we must study only the more general deformed case . the best way to analyze it is to transform the equation ( [ udilatondef ] ) in a schrdinger type .
it is easy to see that an equation of the form @xmath70 can be transformed in a schrdinger type @xmath71 through the transformations @xmath72 with @xmath73 and @xmath74 where the prime is a derivative with respect to @xmath75 . now using the equation ( [ udilatondef ] )
we obtain @xmath76 where @xmath77 , and we take @xmath75 of function of @xmath78 through@xmath79 . therefore , the eingenvalues give us the masses for which the localization is possible .
it is important to note that , after the above transformations , the condition for localization becomes @xmath80 and this is exactly the square integrable condition of the wave function in quantum mechanics .
this enforce the argument that we have a schroedinger - like equation .
despite its importance , the analytic solution for the above equation has not been found .
the strategy therefore is to plot numerically the potential and analyze its shape .
the appearance of resonances depends strongly on the shape of the potential and therefore of the coupling constant .
this plot is given by fig .
[ fig1 ] for @xmath81 .
for this we choose @xmath82 and @xmath83 .
( lined ) scaled by 1/10 , @xmath84 ( dashed ) , @xmath85 ( dotted ) for @xmath2 ( top ) and @xmath3 ( bottom ) . ]
note that the shape of the potential changes significantly and this will give us a very different resonance structure .
another thing we can observe is that the potential becomes null for @xmath86.therefore we do not have a discrete spectrum for @xmath87 . beyond this , it is a known fact from quantum mechanics that any solution for positive @xmath88 must posses a oscillatory contribution , and therefore the wave function is not normalizable .
the conclusion is that the only localized mode is the massless one .
this means that this is the unique mode living inside the membrane .
another important possibility is the appearance of resonances .
as we have a volcano potential , we can ask for the possibility that the massive modes , besides living in the extra dimension , has a peak of probability to be found at the location of the membrane .
this analysis has been done extensively in the literature @xcite .
if we compare our schroedinger equation with that found by these authors , we see that they are basically the same , so that we can extract the same information about resonances of the model .
the authors found that , after normalize the wave function in a truncated region , there is a resonance very close to @xmath44 , and this indicates that as lighter is the mass , bigger is the probability of interacting with the membrane .
them we must corroborate with the analysis and the same happens in our model . from our viewpoint
the wave function has an oscillatory part and can not be normalized .
to scape of this , the authors define a relative probability , but for us it is not clear how it solves the problem .
another very important point is the truncation of the integration region .
this is equivalent to have a potential that do not falls to zero at infinity , therefore it is natural to have a peak of resonance very close to @xmath44 .
this could leads to the wrong interpretation that very light modes can interact with the membrane . from our analysis
, it is not true that as lighter is the massive modes , bigger is the probability of finding them inside the membrane .
in fact , this is corroborated only when @xmath5 and for the value of the coupling constant used by the authors in @xcite . in this case
we find , as can be seen in fig .
[ fig2 ] , a resonance for the specific value @xmath89 .
when we consider @xmath3 , as shown in fig .
[ fig3 ] we find a very rich resonance structure and that heavy modes can also resonate .
we must point here that the schroedinger eq .
used by them is very similar to ours but have a different multiplicative factor in the potential because they study the two form . in principle
this could change the results , therefore in a separate paper we analyze carefully that case @xcite .
in other to analyze resonances , we must compute transmission coefficients(@xmath90 ) , which gives a clearer and cleaner physical interpretation about what happens to a free wave that interact with the membrane .
the idea of the existence of a resonant mode is that for a given mass the potential barrier is transparent to the particle , i. e. , the transmission coefficient has a peak at this mass value .
that means the amplitude of the wave - function has a maximum value at @xmath91 and the probability to find this kk mode inside the membrane is higher . in order to obtain these results numerically ,
we have developed a program to compute transmission coefficients for the given potential profile .
a more extensive and detailed analysis of resonances with transmission coefficients will be given in a separate paper by the authors @xcite . in fig .
[ fig2 ] we give the plots of @xmath90 for @xmath92 and considering the case @xmath93 .
( top ) , @xmath84 ( middle ) , @xmath85 ( bottom ) for @xmath2 . ] as said before , we find resonances for this value of @xmath4 only for @xmath5 . in fig .
[ fig3 ] we show the plot for @xmath3 ( top ) , @xmath84 ( middle ) , @xmath85 ( bottom ) for @xmath3 . ] and we can see how this alter the existence of resonances . for this values
we have an interesting structure of resonances , witch is similar to that found in quantum mechanical problems . as we stress in our separate paper , the existence of resonances depends strongly of the shape of the potential . in our case , therefore , the resonance is driven by the choice of the parameter @xmath4 .
in this paper we have studied the issue of localization for a three form field in a randall - sundrum - like model .
we approach this in a smooth space scenario , where the extra dimension is described by a kink .
we show that , for the usual and the deformed cases , we have the localization of the zero mode only for a specific range of the coupling constant given by @xmath69 .
we study the possibility of resonances in the model .
this is done throughout the numerical computations of the transmission coefficient and we find that the appearance of peaks depends strongly on the value of the dilaton coupling constant . for high values of this constant
we find an interesting structure of resonances , very similar to what is common in quantum mechanical problems .
for the value @xmath3 and @xmath94 , for example , we found eight peaks of resonances .
therefore a bigger value of the coupling constant favor resonances of massive modes .
this should seems natural , as the dilaton is responsible for localizing the zero mode and a bigger value of the constant would augment the possibility for existence of resonances .
it should be noted that from our analysis heavy modes can also resonate .
we thank a. a. moreira for providing a computer for numerical calculations .
the authors would like to acknowledge the financial support provided by fundao cearense de apoio ao desenvolvimento cientfico e tecnolgico ( funcap ) and the conselho nacional de desenvolvimento cientfico e tecnolgico ( cnpq ) .
99 m. nakahara , `` geometry , topology and physics , '' i. oda and s. yahikozawa , `` linking numbers and variational method , '' phys .
b * 238 * , 272 ( 1990 ) .
b. mukhopadhyaya , s. sen , s. sen and s. sengupta , `` bulk kalb - ramond field in randall sundrum scenario , '' phys .
d * 70 * , 066009 ( 2004 ) [ arxiv : hep - th/0403098 ] a. arvanitaki , s. dimopoulos , s. dubovsky , n. kaloper and j. march - russell , arxiv:0905.4720 [ hep - th ] .
p. svrcek and e. witten , jhep * 0606 * , 051 ( 2006 ) [ arxiv : hep - th/0605206 ] .
y. brihaye , t. delsate , n. sawado and y. kodama , [ arxiv : hep - th/1007.0736 ] c. germani and a. kehagias , `` higher - spin fields in braneworlds , '' nucl .
b * 725 * , 15 ( 2005 ) [ arxiv : hep - th/0411269 ] .
a. kehagias and k. tamvakis , `` localized gravitons , gauge bosons and chiral fermions in smooth spaces generated by a bounce , '' phys .
b * 504 * , 38 ( 2001 ) [ arxiv : hep - th/0010112 ] .
l. randall and r. sundrum , `` an alternative to compactification , '' phys .
lett . * 83 * , 4690 ( 1999 ) [ arxiv : hep - th/9906064 ] .
l. randall and r. sundrum , `` a large mass hierarchy from a small extra dimension , '' phys .
* 83 * , 3370 ( 1999 ) [ arxiv : hep - ph/9905221 ] .
b. mukhopadhyaya , s. sen and s. sengupta , `` bulk antisymmetric tensor fields in a randall - sundrum model , '' phys .
d * 76 * , 121501 ( 2007 ) [ arxiv:0709.3428 [ hep - th ] ] .
g. de risi , `` bouncing cosmology from kalb - ramond braneworld , '' phys .
d * 77 * , 044030 ( 2008 ) [ arxiv:0711.3781 [ hep - th ] ]
. b. mukhopadhyaya , s. sen and s. sengupta , `` a randall - sundrum scenario with bulk dilaton and torsion , '' phys .
d * 79 * , 124029 ( 2009 ) [ arxiv:0903.0722 [ hep - th ] ] .
g. alencar , m. o. tahim , r. r. landim , c. r. muniz and r. n. costa filho , `` bulk antisymmetric tensor fields coupled to a dilaton in a randall - sundrum model , '' arxiv:1005.1691 [ hep - th ] .
d. bazeia , a. r. gomes , l. losano and r. menezes , `` braneworld models of scalar fields with generalized dynamics , '' phys .
b * 671 * , 402 ( 2009 ) [ arxiv:0808.1815 [ hep - th ] ] .
d. bazeia , a. r. gomes and l. losano , `` gravity localization on thick branes : a numerical approach , '' int .
j. mod .
a * 24 * , 1135 ( 2009 ) [ arxiv:0708.3530 [ hep - th ] ] .
d. bazeia , f. a. brito and a. r. gomes , `` locally localized gravity and geometric transitions , '' jhep * 0411 * , 070 ( 2004 ) [ arxiv : hep - th/0411088 ] .
d. bazeia , c. furtado and a. r. gomes , `` brane structure from scalar field in warped spacetime , '' jcap * 0402 * , 002 ( 2004 ) [ arxiv : hep - th/0308034 ] .
d. bazeia , l. losano and j. m. c. malbouisson , `` deformed defects , '' phys . rev .
d * 66 * , 101701 ( 2002 ) [ arxiv : hep - th/0209027 ] .
o. dewolfe , d. z. freedman , s. s. gubser and a. karch , `` modeling the fifth dimension with scalars and gravity , '' phys .
d * 62 * , 046008 ( 2000 ) [ arxiv : hep - th/9909134 ] .
a. r. gomes , `` gravity on the bloch brane , '' arxiv : hep - th/0611291 .
w. t. cruz , m. o. tahim and c. a. s. almeida , `` results in kalb - ramond field localization and resonances on deformed branes , '' europhys .
* 88 * , 41001 ( 2009 ) [ arxiv:0912.1029 [ hep - th ] ] .
g. alencar , r. r. landim , m. o. tahim , c. r. muniz and r. n. costa filho , in preparation . | in this article we study the issue of localization of the three - form field in a randall - sundrum - like scenario .
we simulate our membrane by kinks embedded in d=5 , describing the usual case ( not deformed ) and new models coming from a specific deformation procedure .
the gravitational background regarded includes the dilaton contribution .
we show that we can only localize the zero - mode of this field for a specific range of the dilaton coupling , even in the deformed case . a study about resonances
is presented .
we use a numerical approach for calculations of the transmission coefficients associated to the quantum mechanical problem .
this gives a clear description of the physics involved in the model .
we find in this way that the appearance of resonances is strongly dependent on the coupling constant .
we study the cases @xmath0 and @xmath1 for @xmath2 and @xmath3 .
the first value of @xmath4 give us one resonance peak for @xmath5 and no resonances for @xmath6 .
the second value of @xmath4 give us a very rich structure of resonances , with number deppending on the value of @xmath7 .
randall - sundrum , resonances 03.65.-w , 03.65.ca , 03.65.fd , 03.65.ge , 02.30.ik , 02.30.gp |
the quantum volume operator is among the most intensively investigated items in the field of loop quantum gravity and pivotal for the construction of space - time dynamics within this theoretical framework @xcite .
traditionally two versions of such an operator are discussed , due to rovelli and smolin @xcite , and to ashtekar and lewandowski @xcite , respectively , and considerable attention has been devoted to their properties and interrelations@xcite .
more recently , bianchi , dona , and speziale @xcite offered a third proposal for a volume operator which is closer to the concept of spin foams @xcite .
it relies on an older geometric theorem due to minkowski @xcite stating that @xmath0 face areas @xmath1 , @xmath2 , with normal vectors @xmath3 such that @xmath4 for @xmath5 uniquely define a convex polyhedron of @xmath0 faces with areas @xmath1 .
the approach of ref.@xcite amounts in expressing the volume of a classical polyhedron in terms of its face areas , which are in turn promoted to be operators .
minkowskis proof , however , is not constructive , and a remaining obstacle of the above route to a volume operator is to actually find the shape of a general polyhedron given its face areas and face normals @xcite .
such difficulties do not occur in the simplest case of a polyhedron , i.e. a tetrahedron consisting of four faces represented by angular momentum operators coupling to a total spin singlet @xcite .
indeed , for such a quantum tetrahedron all three definitions of the volume operator coincide . on the other hand , for a classical tetrahedron the general phase space parametrization devised by kapovich and millson @xcite results in just one pair of canonical variables , and the ( square of the ) volume operator can explicitly formulated in terms of these quantities @xcite .
moreover , bianchi and haggard have performed a bohr - sommerfeld quantization of the classical tetrahedron where the role of an hamiltonian generating classical orbits is played by the volume operator squared .
the resulting semiclassical eigenvalues agree extremely well with exact numerical data @xcite .
the above observations make clear that classical tetrahedra , arguably the simplest structures a volume can be ascribed to , should be considered as perfectly integrable systems . in turn
, a quantum tetrahedron can be viewed as the `` hydrogen atom '' of quantum spacetime , whereas the next complicated case of a pentahedron might be referred to as the `` helium atom '' @xcite .
most recently , the present author has put forward yet another approach to the semiclassical regime of quantum tetrahedra @xcite . here , by combining observations on
the volume operator squared and its eigenfunctions ( as opposed to the eigenvalues ) , an effective operator in terms of a quantum harmonic oscillator was derived providing an accurate as well as transparent description of the the large - volume sector .
one of the purposes of the present work is to demonstrate the relation between the different treatments of quantum tetrahedra sketched above .
the outline of this paper is as follows . in section [ clpoly ]
we first summarize the kapovich - millson phase space parametrization of general classical polyhedra ( section [ clpolykp ] ) before reviewing and extending in section [ cltetra ] results for the classical tetrahedron .
in particular , we derive an expansion of the volume squared around its maximum and minimum in up to quadrilinear order . section [ quanttetra ] is devoted to the quantum tetrahedron .
we first outline in section [ quanttetragen ] elementary facts about the volume operator and its hilbert space , and we point out several relations between appropriate quantum operators which have analogs in the classical tetrahedron . next the analysis of ref .
@xcite is extended to higher corrections to the resulting harmonic oscillator of up to fourth order .
the results are compared with the outcome of a canonical quantization of the classical volume expression .
the expression of the pentahedral volume in terms of kapovich - millson variables is discussed in appendix [ penta ] .
finally we construct in section [ quantpolygen ] a set of quantum operators for general polyhedra whose commutation relations approach in the semiclassical limit the standard commutators between momentum and position .
let us first recall the essentials of the polyhedral phase space parametrization due to kapovich and millson @xcite .
viewing the vectors @xmath6 as angular momenta , the poisson bracket of arbitrary functions of these variables read @xmath7 in order to implement the closure relation ( [ clorel ] ) one defines @xmath8 for @xmath9 resulting in @xmath10 momenta @xmath11 . defining now @xmath12
such that @xmath13 ( @xmath14 ) and @xmath15 the canonical conjugate variables @xmath16 are then given be the angle between @xmath17 , @xmath18 . indeed , and a straightforward calculation shows that these quantities fulfill indeed the canonical poisson relations @xcite .
@xmath19 the classical volume of a tetrahedron can be expressed as @xmath20 suggesting to investigate the quantity @xmath21 the latter can indeed easily be expressed in terms of the phase space variables @xmath22 , @xmath23 using the observation @xcite @xmath24 moreover , it is easily seen that @xmath25 where @xmath26 is the area of a triangle with edges @xmath27 expressed via herons formula , @xmath28 analogously , using the closure relation ( [ clorel ] ) , @xmath29 such that @xmath30 in order to make closer contact to the quantum tetrahedron to be discussed below , let us introduce the notation @xmath31 fulfilling @xmath32 and @xmath33 with @xmath34 where @xmath35 varies according to @xmath36 with @xmath37 an expression close to ( [ clq ] ) was also found in ref .
@xcite in the semiclassical limit of a quantum tetrahedron .
obviously , @xmath38 is a nonnegative function with @xmath39 , and it is not difficult to verify that it has a unique maximum at some @xmath40 between @xmath41 and @xmath42 @xcite .
thus , @xmath43 has a unique maximum at @xmath44 and @xmath45 while the unique minimum lies at @xmath46 .
expanding around the maximum gives ( @xmath47 , @xmath48 , @xmath49 ) @xmath50 \label{qexpmax}\end{aligned}\ ] ] with @xmath51 and @xmath52 the analogous expansion around the minimum reads ( @xmath53 , @xmath54 ) @xmath55\ , .
\label{qexpmin}\end{aligned}\ ] ] concentrating in both cases on the quadratic contributions , one obtains two harmonic oscillators , @xmath56\ , , \label{oscmax}\\ q_{\rm osc}^{\prime}(p , x ) & = & -\tilde q\left[1-\frac{(p-\pi)^2}{2}-\frac{\tilde\omega^2}{2}x^2\right]\ , .
\label{oscmin}\end{aligned}\ ] ] finally , it is certainly desirable to also express the volume of higher polyhedra in terms of kapovich - millson variables .
appendix [ penta ] details the case of the pentahedron . as shown there
, the above task is certainly feasible , but leads to unpleasantly complicated expressions which inhibit analytical progress .
we begin by reviewing and extending general results of quantum tetrahedra .
a quantum tetrahedron is defined by four angular momentum operators @xmath57 , @xmath58 , representing its faces and coupling to a total singlet@xcite , i.e. the hilbert space consists of all states @xmath59 fulfilling @xmath60 a usual way to construct this space is to couple first the pairs @xmath61 and @xmath62 to two irreducible su(2 ) representations of dimension @xmath63 each . for @xmath61 this standard
construction reads explicitly @xmath64 @xmath65 where @xmath66 are clebsch - gordan coefficients following their usual phase convention@xcite . defining analogous states
@xmath67 for @xmath62 , the quantum number @xmath68 becomes restricted by @xmath69 with @xmath70 the two multiplets @xmath71 , @xmath67 are then coupled to a total singlet , @xmath72 where the phase factor in front will become useful shortly below . the states @xmath59 span a hilbert space of dimension @xmath73 .
the volume operator of a quantum tetrahedron can be formulated as @xmath74 where the operators @xmath75 @xmath58 represent the faces of the tetrahedron with @xmath76 being the planck length squared .
usually the operators @xmath77 are defined with additional prefactors proportional to the immirzi parameter on the r.h.s . of eq .
( [ defe ] ) .
this establishes contact to the general formalism of loop quantum gravity @xcite but is unnecessary for our purposes here . what will become important , however , is that @xmath78 is proportional to @xmath79 . as a result ,
one is led to consider the operator @xmath80 which reads in the basis of the states @xmath59 as @xcite @xmath81 with @xmath82 note the close similarity of the expressions ( [ alpha ] ) and ( [ clbeta ] ) . moreover , in the above basis @xmath83 couples only states with neighboring labels and is represented by a real matrix .
the latter fact depends on the phase factor in the first line of eq .
( [ defsing ] ) .
indeed , upon striping this factor ( which is a unitary operation ) @xmath84 becomes antisymmetric and purely imaginary .
thus , for even @xmath85 , the eigenvalues of @xmath43 come in pairs @xmath86 , and since @xmath87 with @xmath88 , the corresponding eigenstates @xmath89 , @xmath90 fulfill @xmath91 i.e. eigenvectors of eigenvalues differing just in sign are related to each other by changing the sign of any other component . for odd @xmath85
an additional zero eigenvalue occurs @xcite . to make further contact between the classical and the quantum tetrahedron we define in analogy to eqs.([defvw ] ) @xmath92 fulfilling @xmath93 which is the operator analog of eq .
( [ clvxw ] ) .
moreover , one straightforwardly obtains @xmath94 and @xmath95 where @xmath96 is the projector onto the singlet space .
( [ quvsq]),([quwsq ] ) are the operators analogs of eqs .
( [ clvsq]),([clwsq ] ) .
so far we have followed the formalism common to the literature and parametrized the hilbert space of the quantum tetrahedron by a dimensionless quantum number @xmath68 , whereas the phase space variable @xmath35 of the classical tetrahedron has dimension of area . in order to establish closer contact between both descriptions
let us rescale the involved quantum numbers by the planck length squared according to @xmath97 to quantities having also dimension of area . as we shall see below
, this step will also provide a close analogy to standard quantum mechanics in the schrdinger representation .
the analog of the classical expression ( [ classq ] ) reads @xmath98 with @xmath99 the latter quantity shares the essential properties of @xmath38 in eq .
( [ clbeta ] ) .
in particular @xmath100 has a unique maximum at some @xmath101 . in ref .
@xcite the present author has shown how to accurately describe the large - volume ( semiclassical ) regime of @xmath83 ( or @xmath84 ) by a quantum harmonic oscillator in real - space representation with respect to @xmath102 ( or @xmath68 , respectively ) . here
we shall extend this analysis taking into account higher - order corrections within the rescaled variables introduced in the previous section .
let us label the eigenstates of @xmath83 by @xmath103 , @xmath104 , in descending order of eigenvalues with @xmath105 being the state of largest eigenvalue .
with respect to the basis states @xmath59 they can be expressed as @xmath106 thus , taking the view of the standard schrdinger formalism of elementary quantum mechanics , the coefficients @xmath107 are the `` wave function '' of the state @xmath103 with respect to the `` coordinate '' @xmath102 .
the approach of ref .
@xcite starts from evaluating matrix elements @xmath108 between states lying predominantly in the sector of large eigenvalues by approximating the sum by an integral introducing the integration variable @xmath109 , @xmath110 with @xmath111 , @xmath112 .
expanding now @xmath113 around its maximum at @xmath114 and the wave functions @xmath115 , @xmath116 around @xmath117 , one obtains in up to fourth order in the expansions @xmath118 + \frac{c}{3}\ell_p^2\left[\frac{d}{dx},x^3\right ] \biggr]\psi(x ) \label{qmat3}\end{aligned}\ ] ] with @xmath119 , @xmath120 and @xmath121 @xmath122 in calculating the r.h.s . of eq .
( [ qmat3 ] ) we have repeatedly performed integrations by parts and assumed the boundary terms to vanish .
introducing now the operators @xmath123 one easily reads off the effective operator expression @xmath124 + i\frac{c}{3}\left[\hat
p,\hat x^3\right ] \biggr]\ , . \label{qop}\end{aligned}\ ] ] this result extends the findings of ref .
@xcite to higher corrections in the operators @xmath125 , @xmath126 .
the contribution in eq .
( [ qmat3 ] ) involving only derivatives with respect to @xmath117 can be viewed as the result of a continuum approximation according to @xmath127 note also that the symmetric operator ordering in the last term of the second line in eq .
( [ qmat3 ] ) ( i.e. the middle contribution in the second line in eq .
( [ qop ] ) ) emerges from the calculation and not an additional assumption . as a result
, the operator ( [ qop ] ) perfectly matches the classical expression ( [ qexpmax ] ) taking into account the correct operator ordering and the vanishing of the commutators @xmath128=-2i\ell_p^2\hat x\quad,\quad \left[\hat p,\hat x^3\right]=-3i\ell_p^2\hat x^2\ ] ] which are indeed small compared to the other contributions in ( [ qop ] ) as they are proportional to @xmath79 .
alternatively , the matrix elements of such commutators can be viewed to be of higher order in derivatives since @xmath129\psi(x ) \nonumber\\ & & \qquad=\frac{\omega^2}{2 } \int dx\left(\left(\hat p\phi\right)^*\hat x^2\psi -\phi^*\hat x^2\left(\hat p\psi\right)\right)\,,\end{aligned}\ ] ] where the r.h.s contains in total three derivatives with respect to @xmath102 or @xmath117 .
finally the coefficients in the expansions ( [ qop ] ) and ( [ qexpmax ] ) obviously coincide in the limit of large quantum volumes , @xmath130 . in summary ,
up to the commutators discussed above , the operator ( [ qop ] ) is the result of the canonical quantization of the classical expression ( [ qexpmax ] ) via the standard operator replacement ( [ pxop ] ) . when concentrating on the quadratic contributions in eq .
( [ qop ] ) one recovers the harmonic - oscillator expression of ref .
@xcite , @xmath131\label{qosc1}\ ] ] with eigenvalues @xmath132 and corresponding eigenfunctions @xmath133 where @xmath134 are the usual hermite polynomials .
we note that @xmath135 has dimension of inverse area while @xmath136 is dimensionless and can be computed via eqs .
( [ defomega ] ) using @xmath137 given in eq .
( [ alpha ] ) instead of @xmath100 , @xmath138 as stated in ref .
@xcite , the expressions ( [ oscev ] ) and ( [ oscwf ] ) are excellent approximations to the eigenstates and eigenvalues of ( square of the ) the volume operator for already intermediate lengths of the involved spins . this fact is illustrated again in fig .
[ fig1 ] for a typical typical choice of angular momentum quantum numbers all being of order a few ten .
in addition to ref .
@xcite we also plot there the wave function within the lowest - order correction in eq .
( [ qop ] arising from @xmath139 accounted for by first - order perturbation theory .
( filled circles ) for small @xmath140 and a typical choice of angular momentum quantum numbers .
the black solid lines are the unperturbed oscillator wave functions @xmath141 ( in units of @xmath142 ) given in eq .
( [ oscwf ] ) , while the red lines show the eigenfunctions including the first - order perturbation arising form the cubic term @xmath139 in eq .
( [ qop ] ) . ] fig .
[ fig2 ] shows similar data but for smaller spin lengths @xmath143 . here
the oscillator - like features of the wave functions noticeably disappear with increasing @xmath140 , and the corrections from cubic term are clearly more substantial .
( filled circles ) for small @xmath140 and @xmath143 .
the black solid lines are the unperturbed oscillator wave functions @xmath141 ( in units of @xmath142 ) given in eq .
( [ oscwf ] ) , while the red lines show the eigenfunctions including the first - order perturbation arising form the cubic term @xmath139 in eq .
( [ qop ] ) . ] in both figs .
[ fig1 ] and [ fig2 ] we have used the expression @xmath144 as the argument for the wave functions where the additional increment @xmath78 takes into account that @xmath145 couples states of the form @xmath146 and @xmath147 and facilitates comparison with finite size data . with increasing angular momentum quantum numbers , this shift becomes more and more obsolete .
so far we have concentrated on the large and positive eigenvalues of the operator @xmath83 .
the regime of negative eigenvalues of large modulus can be explored by canonically quantizing the classical expression ( [ oscmin ] ) according to the standard recipe ( [ pxop ] ) , @xmath148\ , , \label{qosc2}\ ] ] where we have put for simplicity @xmath149 , @xmath150 .
this operator is related to @xmath151 given in eq .
( [ qosc1 ] ) by a gauge transformation along with a change in sign , @xmath152 such that the eigenfunctions are related by @xmath153 where the phase factor mimics the change in sign stated in the strict relation ( [ signchange ] ) between eigenvectors of @xmath83 to eigenvalues differing in sign only .
in fact , based on this analogy , eq . ( [ oscwf2 ] ) and , as a consequence , eqs .
( [ qosc2]),([qosc3 ] ) have already been given in ref .
here we have provided a more profound derivation based on the canonical quantization of the classical expression ( [ oscmin ] ) . in the operators ( [ pxop ] ) the variable @xmath117 ( and , in turn , @xmath102 )
is considered to be a continuous quantity .
therefore the question remains how to possibly construct a pair of canonical operators retaining the discrete character of @xmath154 with @xmath68 being ( half-)integer . as a step towards this goal
we propose the operators @xmath155 fulfilling @xmath156\nonumber\\ & & \quad=\frac{\ell_p^2}{2i } \sum_{a = a^{\rm min}+\ell_p^2}^{a^{\rm max } } \left(|a\rangle\langle a-\ell_p^2|+|a-\ell_p^2\rangle\langle a|\right)\ , .
\label{pacomm}\end{aligned}\ ] ] for large volumes , the r.h.s .
approaches the unit operator acting on states whose components vary only little on the scale set by @xmath78 , @xmath157\right|\psi\right\rangle & \approx & \sum_{a = a^{\rm min}}^{a^{\rm max}}\langle\phi|a\rangle\langle a|\psi\rangle \label{pacommmat}\\ & = & \frac{\ell_p^2}{i}\langle\phi|\psi\rangle\,.\end{aligned}\ ] ] in fact , the expression ( [ pop ] ) is obviously a discretization of a differential operator .
however , as such discretizations are by no means unique , the question remains open whether there are operators @xmath158 , @xmath159 which ( i ) act on the original discretely - labeled quantum states , ( ii ) turn into @xmath125 , @xmath126 at large volumes , and ( iii ) fulfill @xmath160=\frac{\ell_p^2}{i}\ ] ] as an exact equation on the entire hilbert space
. in full analogy to the kapovich - millson variables we define for a quantum polyhedron of @xmath0 faces ( angular momenta ) the operators @xmath161 for @xmath9 . as the squares of these quantities commute with each other , @xmath162=0\,,\ ] ] orthonormal basis states of the hilbert space can be labeled by quantum numbers @xmath163 according to @xmath164 the closure relation ( [ clorel ] ) translates to @xmath165 i.e. the angular momentum operator @xmath166 couples with the remaining spins @xmath167 , @xmath168 to a total singlet implying @xmath169 with @xmath170 consider now two total singlet states with @xmath171 and @xmath172 , @xmath14 , @xmath173 and all other quantum numbers @xmath174 , @xmath175 identical . then states with @xmath176 ( and all other @xmath174 the same as before )
are also singlets , since @xmath177 and @xmath178 can couple to these values of @xmath163 , and @xmath179 with the above quantum numbers and @xmath180 can couple to the given value of @xmath181 . thus , also the other quantum quantum numbers @xmath163 , @xmath14 , vary within intervals , @xmath182 , and the representation theory of the angular momentum algebra implies @xmath183 with @xmath184 for @xmath185 . without the additional conditions ( [ kcond1]),([kcond2 ] ) the inequalities ( [ kcond3]),([kcond4 ] ) would hold as equalities .
we note , however , that the structure of the quantum numbers @xmath163 is in general quite complex .
for instance , the limiting values @xmath186 , @xmath187 can depend on other quantum numbers @xmath174 , @xmath175 , and the entire structure depends obviously also on the coupling scheme , i.e. the labeling of the operators @xmath188 .
a very simple example is provided by the case @xmath189 in eqs .
( [ tetkcond1]),([tetkcond2 ] ) ; for @xmath190 explicit expressions for @xmath186 , @xmath187 become increasingly tedious . now rescaling the quantum numbers @xmath163 to dimensionful quantities as in section [ quanttetrarescale ] , @xmath191
, we define analogous to eqs .
( [ aop]),([pop ] ) the operators @xmath192 fulfilling the commutation relations @xmath193=\frac{\delta_{ij}\ell_p^2}{2i } \nonumber\\ & & \quad\cdot\sum_{a_1\dots a_{n-3}}\bigl ( & & \quad+|a_1\dots(a_i-\ell_p^2)\dots a_{n-3}\rangle\langle a_1\dots a_{n-3 } r.h.s . approaches the unit operator in just the same way as in eq .
( [ pacomm ] ) .
the investigation of the semiclassical limit of loop quantum gravity is one of the key issues in that approach towards a quantum theory of gravitation . in the present work we have focussed semiclassical properties of quantum polyhedra .
regarding tetrahedra , as their simplest examples , eqs .
( [ defvwop])-([projec ] ) provide operator analogs of classical geometric relations for tetrahedra .
these classical relations are key ingredient to the bohr - sommerfeld analysis of refs .
@xcite .
the expansion of the classical volume squared in up to fourth order in the canonical variables are given in eqs .
( [ qexpmax ] , ( [ qexpmin ] ) .
we have explicitly established the connection between a canonical quantization of these expressions with a recent wave function based approach by the present author @xcite to the large - volume sector of the quantum system . in the leading order
both routes concur yielding a quantum harmonic oscillator as an effective description for the ( square of the ) volume operator . as regards higher orders , the approach of ref .
@xcite leads to additional corrections in terms of commutators which are naturally absent in the classical expressions . including the third - order correction perturbatively leads to improvements of the approximate wave functions .
in fact , it is a distinctive feature of the present work ( and ref .
@xcite ) that it addresses not only the _ eigenvalues _ of the volume operator squared , but also provides very accurate approximations to the _ eigenstates_. furthermore , the comparison of both methods leads also to a full wave function description of the eigenstates of negative eigenvalues of large modulus , a result which could only be conjectured in ref .
@xcite .
differently from previous formulations , the position variable used here is chosen to have dimension of ( planck ) length squared , @xmath76 .
this definitional detail is by no means necessary but facilitates the identification of quantum corrections .
the ultimate reason for the latter observation is the fact that plancks constant @xmath79 itself is dimensionful .
a further interesting point is the zero eigenvalue occurring for tetrahedra with odd hilbert space dimension @xmath85 where the eigenstate can be given , up to normalization , in a closed form @xcite .
moreover , the bohr - sommerfeld quantization carried out by bianchi and haggard @xcite yields also surprisingly accurate results for eigenvalues of such small modulus .
thus the question arises whether the eigenstates corresponding to zero eigenvalues can also be cast , for large angular momenta @xmath194 , in a wave function of a continuous variable .
an important step towards extending the results for the tetrahedron to higher polyhedra is to express their volume in terms of canonical variables . in appendix [ penta ]
we have achieved this goal for the case of pentahedra .
however , the resulting expressions are particularly lengthy and complex such that further practical progress seems to require dedicated numerics and/or extensive but judicious use of computer algebra , which is beyond the scope of the present investigation . for general quantum polyhedra described by discrete angular momentum quantum numbers
we have formulated a set of quantum operators fulfilling in the semiclassical regime the standard commutation relations between momentum and position .
indeed a major challenge is of course the analysis of the volume operator(s ) for higher polyhedra .
results towards this goal were obtained in refs .
@xcite for pentahedra , and for the general case the operators constructed in the present paper in analogy to the kapovich - millson variables of the classical phase space might , although fairly straightforward , provide a useful step .
yet another possible route for generalizing the present investigations is to study polytopes in higher dimensions which also allow for a description in terms of su(2 ) intertwiners @xcite .
in this appendix we discuss the volume of a classical pentahedron in terms of kapovich - millson variables .
we concentrate on the dominant type of pentahedra which define the submanifold of maximal dimension in the phase space of the variables @xmath6 @xcite . these pentahedra are trigonal prisms whereas a subdomiant type is given by pyramidal pentahedra .
these can be generated form the former type by collapsing an edge connecting the two trigonal faces onto a single point and form therefore a submanifold of lower dimension .
an expression for the volume of a trigonal prism has been devised by haggard @xcite starting from the observation that such an object can always be extended to a tetrahedron . using the labeling on the left of fig .
[ fig3 ] this extended body fulfills the closure relation @xmath195 and by projecting this equation onto appropriate cross products the above coefficients are easily obtained as @xmath196 with @xmath197 . the volume can now be expressed as the difference ot two tetrahedral volumes , @xmath198 our goal is now to express the scaling coefficients ( [ coeff1 ] ) occuring along @xmath199 in eq .
( [ pentvol ] ) in terms of standard kapovich - millson variables .
according to the presrciption given in section [ clpolykp ] we define @xmath200 and @xmath201 such that the variables @xmath16 , @xmath202 , conjugate to @xmath203 are the angles between @xmath17 , @xmath18 . using the definition ( [ defdel ] ) one has @xmath204 moreover , the relations @xmath205 allow us to achieve a part of our task in a comparatively compact manner , @xmath206 unfortunately , the remaining quantities entering the volume ( [ pentvol ] ) will turn out to lead to clearly lengthier expressions . to compute them ,
we shall not aim at accessing further triple products @xmath207 directly but rather project the closure relation ( [ scaling1 ] ) onto @xmath208 , @xmath209 providing two further equations for @xmath210 , @xmath211 , @xmath212 with coeffcients we will determine now . in order to determine @xmath216 we calculate , using eq .
( [ p1a1 ] ) , @xmath217 such that , taking into account q. ( [ p1a3 ] ) @xmath218 and finally @xmath219 where @xmath220 similarly , one finds @xmath221 which yields in combination with eq .
( [ p2a4 ] ) @xmath222 with @xmath223 thus we have expressed all scalar products occurring in eqs .
( [ prop1]),([prop2 ] ) in terms of the canonical variables @xmath224 , @xmath16 . taking into account eq .
( [ b - a ] ) , these relations can now be formulated as @xmath225 with @xmath226 and @xmath227\nonumber\\ & & \quad+p_2 ^ 2+a_4 ^ 2-a_5 ^ 2\,.\end{aligned}\ ] ] now , inverting the @xmath228-matrix ( [ defm ] ) and using again eq .
( [ b - a ] ) one can explicitly solve for the scaling coefficients @xmath210 , @xmath211 , @xmath212 .
this procedure , however , will obviously result in forbiddingly lengthy and complicated expressions , which is mainly due to the cumbersome structures in on the r.h.s of eq .
( [ a+bg ] ) .
we note that these expressions considerably simplify for @xmath229 and @xmath230 , i.e. if the two traingular faces and two of the other faces have pairwise the same area .
then one has @xmath231 however , also these quantities seem to be too complicated to allow for further practical analytical progress towards , e.g. , the extrema of the pentahedral volume and expansions around them
. one might suspect that simpler expressions for the pentahedral volume can be obtained by a judicious alternative choice of the canonical variables .
for example , an apparently more symmetric arrangement would be to couple the trigonal faces separately with other faces to canonical momenta . using the labeling given on the right of fig .
[ fig3 ] , this means to use the definitions ( [ defp1 ] ) , ( [ defv1]),([defw1 ] ) as before and put @xmath232 however , the closure relation ( [ clorel ] ) immediately tells that @xmath233 thus , up to inessential signs , we end up with the same canonical variables as before .
moreover , the closure realtion for the extended tetrahedral volume reads now @xmath234 where the new scaling coefficients can be expressed in terms of the old ones ( [ coeff1 ] ) as @xcite @xmath235 as a result , we encounter very similar technical difficulties .
furthermore , in ref .
@xcite an exhaustive list of pentahedral face labelings and corresponding scaling coefficients has been given .
inspecting these results does also not give rise to the hope that such a change of variables will lead to substantially simpler expressions for the volume . | quantum polyhedra constructed from angular momentum operators are the building blocks of space in its quantum description as advocated by loop quantum gravity . here
we extend previous results on the semiclassical properties of quantum polyhedra . regarding tetrahedra
, we compare the results from a canonical quantization of the classical system with a recent wave function based approach to the large - volume sector of the quantum system .
both methods agree in the leading order of the resulting effective operator ( given by an harmonic oscillator ) , while minor differences occur in higher corrections .
perturbative inclusion of such corrections improves the approximation to the eigenstates .
moreover , the comparison of both methods leads also to a full wave function description of the eigenstates of the ( square of the ) volume operator at negative eigenvalues of large modulus . for the case of general quantum polyhedra described by discrete angular momentum quantum numbers we formulate a set of quantum operators fulfilling in the semiclassical regime the standard commutation relations between momentum and position . differently from previous formulations ,
the position variable here is chosen to have dimension of ( planck ) length squared which facilitates the identification of quantum corrections .
finally , we provide expressions for the pentahedral volume in terms of kapovich - millson variables . |
a beautiful consequence of mike freedman s disk embedding theorem is the existence of non - smoothable @xmath0-manifolds . in the easiest
setting , his result can be stated as follows .
[ thm : classification ] any odd unimodular symmetric form @xmath1 is realized as the intersection form of exactly _ two _ closed simply - connected oriented 4manifolds ( up to homeomorphism ) .
these 4manifolds are homotopy equivalent and are distinguished by the following ( equivalent ) criteria : exactly one of the manifolds 1 .
is smoothable after crossing with @xmath2 .
is smoothable after connected sum with finitely many copies of @xmath3 .
3 . has a linear reduction of its micro normal bundle .
has vanishing kirby - siebenmann invariant in @xmath4 .
exhibits the following formula for a quadratic refinement @xmath5 of @xmath6 : @xmath7 by donaldson s theorem a @xcite , _ exactly _ the diagonalizable odd forms @xmath6 are realized by closed _ smooth _
diagonal forms are realized by connected sums of complex projective planes ( with varying orientations ) ; in fact , most such forms are now known to admit infinitely many smooth representatives ( all being homeomorphic by the above theorem ) , see e.g. @xcite .
criterion ( v ) is the most elementary and will be explained in detail in section [ sec : ks ] .
the key is the following geometric interpretation for the quadratic refinement @xmath8 . in a simply - connected closed 4manifold @xmath9 , any class in @xmath10
can be represented by a ( topologically generic ) immersed sphere @xmath11 .
this means that @xmath12 looks locally like @xmath13 , except for finitely many double points around which @xmath12 looks like @xmath14 .
one can add more local self - intersection points to @xmath12 until their algebraic sum is zero .
this implies that one can choose _ whitney disks _
@xmath15 , pairing all these self - intersection points . these are ( topologically generic ) immersed disks
@xmath16 whose boundary consists of two arcs , each going between the two intersection points but on different sheets , see figure [ fig : whitney - move ] .
we will explain in lemma [ lem : c ] why @xmath8 equals an intersection invariant @xmath17 , computed by summing the ( topologically generic ) intersections between an immersed sphere @xmath12 , representing the characteristic @xmath18 , and ( the interiors of ) framed whitney disks @xmath15 for @xmath12 : @xmath19 in @xcite , this invariant was called the _ kervaire - milnor invariant _ because these authors first proved rohlin s formula below @xcite for the case where @xmath9 is smooth and @xmath20 is represented by an embedded sphere , implying the properties @xmath21 .
[ rem : framing ] the figure above shows a _ framed _ whitney disk @xmath15 in the sense that there are two _ disjoint _ parallel copies of @xmath15 , as needed for the whitney move on the right hand side . in general , a whitney disk comes with a framing of its boundary and hence admits a well defined euler number in @xmath22 , its _
twist_. the operation of _ boundary twisting _
@xcite allows to assume that all whitney disks are framed , i.e. have twist zero .
moreover , one can also assume that the @xmath15 are ( disjointly ) embedded disks , by pushing all ( self)-intersections off the boundary .
a generalization of rohlin s theorem @xcite says that this geometric invariant determines the kirby - siebenmann invariant of a closed oriented 4manifold @xmath9 by the formula @xmath23 explaining the equivalence of criteria ( iv ) and ( v ) above . in section [ sec : ks ] we ll recall a definition of @xmath24 which makes the above formula hold for all closed oriented 4manifolds @xmath9 ( without assuming that @xmath20 is spherical ) .
the 2-complex @xmath25 in @xmath9 is referred to as a _ whitney tower _ of order 1 , built on @xmath12 , with order 1 whitney disks @xmath15 .
the invariant @xmath26 used above is the first _ intersection invariant _ of such whitney towers .
it has again _ order 1 _ , the order zero intersection invariants being given by the intersection form @xmath27 . in a sequence of papers , the authors generalized this invariant to higher orders , see for example our survey @xcite .
the idea is that if @xmath28 vanishes then all intersections between @xmath12 and @xmath15 can be paired by _ order 2 _ whitney disks @xmath29 and there should be a second - order intersection invariant @xmath30 measuring the obstruction for finding order 3 whitney disks , and so on . in @xcite
we worked out this higher - order intersection theory in detail for whitney towers built on immersed disks in the @xmath0ball bounded by framed links in the @xmath31sphere . in this
simply connected setting the invariant @xmath32 of an order @xmath33 ( framed ) whitney tower @xmath34 takes values in an abelian group @xmath35 ( where @xmath36 is number of link components ) , and the vanishing of @xmath32 implies that the link bounds and order @xmath37 whitney tower . for links bounding _ twisted
_ whitney towers there is an analogous obstruction theory and intersection invariant @xmath38{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } _
k({\mathcal{w}})\in{\mathcal{t}}^{{\mathchoice { \begin{minipage}{.15in}\includegraphics[width=.15in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } _ k(m)$ ] , and in the second part of this paper we develop an algebraic theory of quadratic forms , leading to a beautiful relation between these framed and twisted obstruction groups , spelled out in theorem [ thm : exact ] .
this result is used in the computation of the whitney tower filtration on classical links described in @xcite .
* acknowledgments : * the main part of this paper was written while the first two authors were visiting the third author at the max - planck - institut fr mathematik in bonn .
they all thank mpim for its stimulating research environment and generous support .
the third author was also supported by nsf grants dms-0806052 and dms-0757312 .
in the notation of the introduction , recall that a class @xmath39 is called _ characteristic _ if @xmath40 the set @xmath41 of characteristic elements is a @xmath42-torsor via the action @xmath43 .
a closed oriented 4manifold @xmath9 with @xmath44 defines a _
quadratic refinement _
@xmath45 of @xmath6 in the sense that @xmath46 this means that @xmath24 is completely determined by _ one _ of its values ( and that value is determined by the kirby - siebenmann invariant of @xmath9 via rohlin s theorem above ) .
if @xmath6 is even then @xmath47 is characteristic and a simple geometric argument shows that @xmath48 , see below .
that s why @xmath24 is only interesting for odd intersection forms . freedman and kirby @xcite defined @xmath49 by representing @xmath20 by an embedded surface @xmath50 such that @xmath51 has a spin structure that does not extend across @xmath52 .
let @xmath53 be the projection map of the boundary of a normal disk bundle for @xmath52 .
then there is a quadratic refinement @xmath54 of the intersection form on @xmath55 defined by taking the inverse image torus @xmath56 for a circle @xmath57 in @xmath52 and observing that it comes equipped with a canonical spin structure .
then @xmath58 is the arf invariant of this spin torus , aka its spin bordism class , and finally @xmath49 is the arf invariant of @xmath54
. this definition can be drastically simplified if @xmath20 is represented by an immersed sphere @xmath12 whose self - intersection points are paired by framed whitney disks @xmath59 .
( this is always the case if @xmath9 is simply - connected as in the introduction . )
[ lem : c ] define @xmath60 to be the sum of all intersections between the immersed sphere @xmath12 and the interiors of the whitney disks @xmath15
. then @xmath61 . in @xcite the following simplification of @xmath49
is already explained , in fact , this was the original definition ( and only later it was realized that one needs a more general approach because caps do nt always exist ) : assume that @xmath20 is represented by a surface @xmath52 with ( immersed , framed ) _
caps_. these are ( immersed , framed ) disks @xmath62 in @xmath9 bounding a hyperbolic basis @xmath63 of curves in @xmath55 .
freedman - kirby show that the quadratic form @xmath64 is determined by @xmath65 number of intersections between the interior of the cap @xmath66 and @xmath52 , similarly for @xmath67 . by definition of the arf invariant
, one gets that @xmath68 assume now that @xmath20 is represented by an immersed sphere @xmath12 whose self - intersection points are paired by ( immersed , framed ) whitney disks @xmath69 .
we can get into the capped surface situation as follows : for each pair of self - intersection points of @xmath12 , add a tube @xmath70 on one sheet going from one self - intersection to the other . that turns @xmath12 into an embedded surface @xmath52 with half of the caps @xmath66 given by small normal disks to @xmath52 that bound the generating circles on @xmath70 . moreover
, the whitney disks @xmath15 can serve as the dual caps @xmath71 , preserving the framing , as illustrated in figure [ fig : framings - preserved ] . by construction , @xmath72 since each normal disk @xmath66 intersects @xmath52 in a single point .
therefore , the required formula follows : @xmath73 [ rem : tau well - defined ] in the simply - connected case , it is not hard to see that @xmath74 is well - defined exactly on characteristic elements .
one thing to check is that it does not depend on the choices of the whitney disks @xmath15 .
once we fix the boundary , any two such choices differ by a connected sum into a sphere @xmath75 . if we require the whitney disks to be ( stably ) framed then @xmath75 needs to be ( stably ) framed and hence it intersects a characteristic sphere in an even number of points , leaving our count @xmath76 unchanged modulo two .
all these considerations can be found in chapter 10 of the book @xcite by freedman and quinn .
unfortunately , the results do nt hold as stated for 4manifolds with fundamental groups that contain 2-torsion elements .
the problem arises from different choices of pairings of intersection points , as observed by richard stong in @xcite . in @xcite
, the last two authors gave a complete discussion of the invariant @xmath76 in the presence of fundamental groups .
all trees considered in this paper are _ unitrivalent , oriented and labelled_. this means that they are equipped with _
vertex orientations _ , i.e. cyclic orderings of the edges incident to each trivalent vertex
. moreover , the univalent vertices of a tree are labeled by elements of the index set @xmath77 , except for the unlabeled _ root vertex _ if a given tree happens to be rooted .
( there is at most one root , even though any other index can appear several times ) .
all trees are considered up to label - preserving isomorphism .
the _ order _ of such a tree is the number of trivalent vertices . given rooted trees @xmath78 and @xmath79 , the _ rooted product _
@xmath80 is the rooted tree gotten by identifying the two roots to a vertex and adjoining a rooted edge to this new vertex , with the orientation of the new trivalent vertex given by the ordering of @xmath78 and @xmath79 in @xmath80 . the _ inner product _
@xmath81 of two rooted trees @xmath78 and @xmath79 is defined to be the unrooted tree gotten by identifying the two rooted edges to a single edge .
we observe that the two products interact well in the sense of figure [ fig : rotate ] .
let @xmath82 be the free abelian group generated by ( isomorphism classes of ) rooted trees as above .
it is graded by order and the rooted product can be extended linearly to a pairing @xmath83 this is grading preserving on @xmath84 $ ] , i.e. it preserves the grading when shifted up by one ( so order is replaced by the number of univalent non - root vertices ) . on the other hand , the inner product @xmath85 is grading preserving via order . here
@xmath86 is the free abelian group generated by unrooted trees as above .
note that rotating the relevant planar trees by 180 respectively 120 degrees shows that the inner product is both _
symmetric _ and _ invariant _ : @xmath87 and @xmath88 , see figure [ fig : rotate ] for the proof of invariance .
[ def : tree - groups ] the graded abelian groups @xmath89 respectively @xmath90 are defined as quotients of @xmath82 respectively @xmath86 by the as and ihx relations as in figure [ fig : relations ] . and @xmath90 .
all trivalent orientations come from an orientation of the plane , and univalent vertices extend to subtrees which are fixed in each equation . ]
it is well known that @xmath89 is the free ( quasi ) lie algebra over @xmath22 on @xmath36 generators with lie bracket induced by the rooted product . here the word _
_ refers to the fact that we only require the antisymmetry relations @xmath91= - [ y , x]$ ] .
so @xmath92 $ ] is not necessarily zero in these lie algebras . in our previous papers , we needed to consider both versions of lie algebras and used the notation @xmath93 for @xmath94 ( recall that one gets a _ graded _ lie algebra only when shifting the order by one ) . in this paper
we will only study one type of lie algebras and usually omit the adjective ` quasi ' .
the inner product extends uniquely to a bilinear , symmetric , invariant pairing @xmath95 this follows simply from observing that the as and ihx relations hold on both sides and are preserved by the inner product .
we will show in lemma [ lem : inner product ] that this inner product is in fact _
universal_. [ def : twisted - tree - groups ] the group @xmath96{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } } _ { 2n}(m)$ ] is gotten from @xmath97 by adding order @xmath33 @xmath98{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } $ ] -trees as new generators . these are rooted trees of order @xmath33 as above , except that the root carries the label @xmath98{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } $ ] .
in addition to the ihx- and as - relations on unrooted trees in @xmath97 , we introduce the following new _ symmetry _ , _ interior twist _ and _ twisted ihx _ relations : @xmath99{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } = ( -j)^{{\mathchoice { \begin{minipage}{.15in}\includegraphics[width=.15in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } \quad\quad 2\cdot j^{{\mathchoice { \begin{minipage}{.15in}\includegraphics[width=.15in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } = \langle j , j\rangle \quad\quad i^{{\mathchoice { \begin{minipage}{.15in}\includegraphics[width=.15in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } = h^{{\mathchoice { \begin{minipage}{.15in}\includegraphics[width=.15in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } + x^{{\mathchoice { \begin{minipage}{.15in}\includegraphics[width=.15in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } - \langle h , x\rangle\ ] ] as their names suggest , these new relations arose from geometric considerations for twisted whitney towers in @xcite .
they will be explained algebraically in our last section via the theory of universal quadratic refinements .
roughly speaking , the universal symmetric pairing @xmath100 will be shown to admit a universal quadratic refinement @xmath101{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } ( m)$ ] defined by @xmath102{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } $ ] .
in particular , with the right algebraic notion of ` quadratic refinement ' , the group @xmath103{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } ( m)$ ] is completely determined by the pairing @xmath100 .
the rest of this paper is devoted to finding this notion . as a consequence
, we will prove the following exact sequence at the very end of this paper .
it was used substantially in @xcite for the classification of whitney towers in the 4-ball .
[ thm : exact ] for all @xmath104 , the maps @xmath105 respectively @xmath106{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } \mapsto 1\otimes j$ ] give an exact sequence : @xmath107{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } _ { 2n}(m ) { \longrightarrow}{\mathbb{z}}_2 \otimes { \mathcal{l}}_{n}(m ) { \longrightarrow}0\ ] ]
in this section we explain an algebraic framework into which our groups @xmath90 and @xmath108{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } ( m)$ ] fit naturally . in lemma
[ lem : inner product ] we show that the @xmath90-valued inner product @xmath109 on the free lie algebra is universal .
then a general theory of quadratic refinements is developed and specialized from the non - commutative to the commutative to finally , symmetric settings . in corollary
[ cor : quadratic ] we show that @xmath103{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } ( m)$ ] is the home for the universal quadratic refinement of the @xmath97-valued inner product @xmath110 .
we work over the ground ring of integers but all our arguments go through for any commutative ring .
we also only discuss the case of finite generating sets @xmath111 , even though everything holds in the infinite case .
[ lem : inner product ] let @xmath112 be a lie algebra together with a bilinear , symmetric , invariant pairing @xmath113 into some abelian group @xmath9 .
if @xmath114 is a lie homomorphism ( given by @xmath36 arbitrary elements in @xmath112 ) there exists a _
unique _ linear map @xmath115 such that for all @xmath116 @xmath117 the uniqueness of @xmath118 is clear since the inner product map is onto . for existence
, we first construct a map @xmath119 as follows . given a tree @xmath120 pick an edge in @xmath121 to split @xmath122 for rooted trees @xmath123 .
then set @xmath124 if we split @xmath121 at an adjacent edge , this expression stays unchanged because of the symmetry and invariance of @xmath6 .
however , one can go from any given edge to any other by a sequence of adjacent edges , showing that @xmath125 does not depend on the choice of splitting .
it is clear that @xmath126 can be extended linearly to the free abelian group on @xmath86 and since @xmath127 preserves as and ihx relations by assumption , this extension factors through a map @xmath118 as required .
[ rem : grading ] recall that @xmath128 $ ] is actually a graded lie algebra , i.e. the lie bracket preserves the grading when shifted up by one ( so order is replaced by the number of univalent non - root vertices ) .
let s assume in the above lemma that the groups @xmath129 are @xmath22-graded , @xmath130 $ ] is a graded lie algebra and that @xmath131 preserve those gradings .
then the proof shows that the resulting linear map @xmath118 also preserves the grading .
the rest of this section describes a general setting for relating our groups @xmath108{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } _ { 2n}(m)$ ] that measure the intersection invariant of twisted whitney towers to a universal ( symmetric ) quadratic refinement of the @xmath97-valued inner product .
we first give a couple of definitions that generalize those introduced by hans baues in @xcite and @xcite , and ranicki in @xcite .
these will lead to the most general notion of quadratic refinements for which we construct a universal example .
later we shall specialize the definitions from _ non - commutative _ to _ commutative _ and finally , to _ symmetric _ quadratic forms and construct universal examples in all cases .
@xmath139 will serve as the range of the ( non - commutative ) quadratic forms defined below
. we will write both groups additively since in most examples @xmath137 turns out to be commutative .
a morphism @xmath140 between quadratic groups is a pair of homomorphisms @xmath141 such that both diagrams involving @xmath142 commute .
[ ex : m ] the example motivating the notation comes from homotopy theory , see e.g.@xcite . for @xmath143 ,
let @xmath144 , @xmath145 , @xmath146 be the hopf invariant and @xmath136 be given by post - composing with the whitehead product @xmath147:s^{2n-1}\to s^n$ ] .
this quadratic group satisfies @xmath148 which is part of the definition used in @xcite , where @xmath137 is also assumed to be commutative .
as we shall see , these additional assumptions have the disadvantage that they are not satisfied for the universal example [ ex : universal ] .
another important example comes from an abelian group with involution @xmath149
. then we let @xmath150):=x+x^*\ ] ] and @xmath136 be the natural quotient map .
for example , if @xmath9 is a ring with involution @xmath151 , then we get two possible involutions on the abelian group @xmath9 : @xmath152 .
the choice of sign determines whether we study symmetric respectively skew - symmetric pairings .
[ lem : involutions ] given a quadratic group , the endomorphism @xmath155 gives an involution on @xmath135 ( which we will denote by @xmath156 ) .
moreover , the formula @xmath157 defines an anti - involution on @xmath137 .
these satisfy the proof of lemma [ lem : involutions ] is straightforward and will be left to the reader . to show that @xmath161 is an anti - homomorphism one uses that @xmath162 is central and that @xmath163 is an anti - homomorphism .
[ def : quadratic group refinement ] a quadratic group @xmath139 is a _
quadratic refinement _ of an abelian group with involution @xmath149 if @xmath164 it follows from ( i ) in lemma [ lem : involutions ] that in this case , the image of @xmath146 lies in the fixed point set of the involution : @xmath165 .
it follows from ( ii ) in lemma [ lem : involutions ] that the additional condition @xmath148 used in @xcite is satisfied if and only if @xmath167 , or equivalently , if @xmath136 factors through the cofixed point set of the involution : @xmath168 it follows that the notion in @xcite is equivalent to that in @xcite , except that @xmath135 is assumed to be the ground ring @xmath169 in the former . in that case ,
our involution is simply @xmath170 , where @xmath171 and @xmath151 is the given involution on the ring @xmath169 .
then @xmath172-symmetric forms in the sense of ranicki become hermitian forms in the sense defined below . in particular , ranicki s @xmath173-symmetric forms are different from the notion of _ symmetric _ form in this paper : we reserve it for the easiest case where both involutions , @xmath156 and @xmath161 , are trivial .
[ def : quadratic ] a _ ( non - commutative ) quadratic form _ on an abelian group @xmath174 with values in a ( non - commutative ) quadratic group @xmath175 is given by a bilinear map @xmath176 and a map @xmath177 satisfying we say that @xmath180 is a _
quadratic refinement _ of @xmath6 : property ( i ) says that @xmath180 is quadratic and property ( ii ) means that it `` refines '' @xmath6 .
the notation @xmath137 and @xmath135 was designed ( by baues ) to reflect the number of variables ( * e*ntries ) of the maps @xmath180 and @xmath6 respectively .
he also writes @xmath181 and @xmath182 , however , we decided not to follow that part of the notation .
we write @xmath183 for such quadratic forms and we always assume that the quadratic group @xmath139 is part of the data for @xmath184 .
this means that the morphisms in the category of quadratic forms are pairs of morphisms @xmath185 such that both diagrams involving @xmath186 commute .
[ lem : symmetries ] let @xmath183 be a quadratic form as above .
then @xmath6 is _ hermitian _ with respect to the involution @xmath187 on @xmath135 : @xmath188 and @xmath180 is _ hermitian _ with respect to the anti - involution @xmath189 on @xmath137 : @xmath190 starting with a hermitian form @xmath6 with values in a group with involution @xmath149 , the first step in finding a quadratic refinement for @xmath6 is to find a quadratic refinement @xmath139 of @xmath149 in the sense of definition [ def : quadratic group refinement ] , motivating our terminology .
[ ex : universal ] given a hermitian form @xmath195 , one gets a quadratic refinement @xmath196 of @xmath6 as follows . set @xmath197 and define the universal target @xmath198 to be the group consisting of pairs @xmath199 with @xmath200 and @xmath201 and multiplication given by @xmath202 in other words , @xmath203 is the central extension @xmath204 & m_{ee } \ar[r ] & m_{ee } \times_\lambda a \ar[r ] & a \ar[r ] & 1 } \ ] ] determined by the cocycle @xmath6 , compare section [ sec : presentations ] .
it follows that @xmath137 is commutative if and only if @xmath6 is _ symmetric _ in the naive sense that @xmath205 . set @xmath206
we claim that @xmath207 is a quadratic group as in definition [ def : quadratic group ] .
it is clear that @xmath208 is a homomorphism with image in the center of @xmath137 .
the homomorphism property of @xmath209 follows from the fact that @xmath6 is bilinear and hermitian : @xmath210 condition ( iii ) of a quadratic group is also checked easily : @xmath211 we also see that @xmath212 which means that @xmath213 `` refines '' ( in the sense of definition [ def : quadratic group refinement ] ) the group with involution @xmath149 .
finally , setting @xmath214 , we claim that @xmath215 is a quadratic refinement of @xmath6 .
we need to check properties ( i ) and ( ii ) of a quadratic form ( definition [ def : quadratic ] ) : ( i ) is simply @xmath216 , and ( ii ) explains why we used a sign in front of @xmath6 in our central extension : @xmath217 the following result will show that @xmath196 is indeed a _ universal _ quadratic refinement of @xmath6 .
this is the content of the first statement in the theorem below .
it follows from the second statement because for any quadratic refinement @xmath180 of @xmath6 it shows that forgetting the quadratic data gives canonical isomorphisms @xmath218 where @xmath219
respectively @xmath220 are ( the morphisms in ) the categories of quadratic respectively hermitian forms . since @xmath221 and the morphisms in the category @xmath222 of quadratic refinements of @xmath6 by definition all lie over the identity of @xmath6 , the set @xmath223 contains a unique element , namely the required universal morphism @xmath224 .
[ thm : universal ] the quadratic form @xmath225 is initial in the category of quadratic refinements of @xmath6 . in fact , the forgetful functor @xmath226 from the category of quadratic forms to the category of hermitian forms has a left adjoint @xmath227 given by @xmath228 .
we have to construct natural isomorphisms @xmath229 for any quadratic form @xmath230 and hermitian form @xmath6 .
recall that the morphisms in @xmath219 are pairs @xmath231 and @xmath232 such that the relevant diagrams commute .
this implies that forgetting about the quadratic datum @xmath233 gives a natural map from the left to the right hand side above .
given a morphism @xmath234 consisting of homomorphisms @xmath235 and @xmath236 such that @xmath237 we need to show that there is a _
homomorphism @xmath238 such that the following 3 diagrams commute : @xmath239
|{(1 ) } m_{e } \ar[r]^{h } \ar[d]_{\beta_e } & m_{ee } \ar[d]^{\beta_{ee } } & \ar @ { } [ dr ] |{(2 ) } m_{ee } \ar[r]^{p } \ar[d]_{\beta_{ee } } & m_{e } \ar[d]^{\beta_{e } } & \ar @ { } [ dr ] |{(3 ) } a \ar[r]^{\mu_{\lambda } } \ar[d]_{\alpha } & m_{e } \ar[d]^{\beta_{e } } \\
m_{e } ' \ar[r]^{h ' } & m_{ee } ' & m_{ee } ' \ar[r]^{p ' } & m_{e } ' & a ' \ar[r]^{\mu ' } & m_{e } ' } \ ] ] we will now make use of the fact that @xmath240 because @xmath196 is given as in example [ ex : universal ] . in this case , diagrams ( 2 ) and ( 3 ) are equivalent to @xmath241 because @xmath242 and @xmath243 .
this implies directly the uniqueness of @xmath233 . for existence
, we only have to check that the formula @xmath244 gives indeed a group homomorphism @xmath245 that makes diagram ( 1 ) commute .
note that the image of @xmath246 is central in @xmath247 and hence the order of the summands does not matter .
we have @xmath248 to get to the fourth line , we used property ( ii ) of a quadratic form to cancel the term @xmath249 . for the commutativity of diagram ( 1 ) we use property ( i ) of a quadratic form , as well as the fact that @xmath250 preserves the involution @xmath156 :
@xmath251 this finishes the proof of left adjointness of @xmath227 . if the bilinear form @xmath6 happens to be _ symmetric _ , or more precisely , if it takes values in a group @xmath135 with _ trivial _ involution @xmath156 , then the above construction still gives a quadratic refinement @xmath196 .
its target quadratic group @xmath213 has the properties that @xmath137 is abelian and @xmath252 .
it is not hard to see that our construction above leads to the following result .
[ thm : universal ] for any symmetric form @xmath6 one can functorially construct a quadratic form @xmath225 that is initial in the category of quadratic refinements of @xmath6 with trivial involution @xmath156 .
in fact , the forgetful functor @xmath226 from the category of quadratic forms with trivial involution @xmath156 to the category of symmetric forms has a left adjoint @xmath253 .
the case where @xmath156 is non - trivial but the anti - involution @xmath161 on @xmath137 is trivial is even more interesting . in this case
, @xmath6 is still hermitian with respect to @xmath156 but one is only interested in quadratic refinements @xmath180 that are symmetric in the sense that @xmath257 .
this case deserves its own definition : in fact , a commutative quadratic group is the same thing as a non - commutative quadratic group with trivial anti - involution @xmath161 .
this comes from the fact that the squaring map @xmath259 is a homomorphism if and only if @xmath137 is commutative .
our universal example @xmath213 is in general _ not _ commutative because one gets in this case @xmath260 however , we shall see in theorem [ thm : commutative universal ] that we can just divide by these relations @xmath261 to obtain another universal quadratic refinement of a given hermitian form @xmath6 but this time with values in a _ commutative _ quadratic group .
before we work this out , let us mention the essential example from topology .
consider a manifold @xmath262 of dimension @xmath263 and let @xmath139 be as in ( m ) from example [ ex : m ] with @xmath264 $ ] .
in particular , we have @xmath265 but in general the involution @xmath156 is non - trivial . on group elements ,
it is given by @xmath266 with @xmath267 ( induced by ) the first stiefel - whitney class of @xmath262 .
then the equivariant intersection form @xmath268 on @xmath269 is bilinear and hermitian as required .
moreover , the self - intersection invariant @xmath270 defined by wall @xcite gives a quadratic refinement of @xmath271 , at least on the subgroup @xmath174 of elements represented by immersed @xmath33-spheres with vanishing normal euler number .
since the involution @xmath274 is trivial by assumption , we already know that @xmath257 from lemma [ lem : symmetries ] .
thus it suffices to prove the claim for positive @xmath275 by induction : @xmath276 here we again used the fact that @xmath277 .
[ thm : commutative universal ] any hermitian bilinear form @xmath6 has a universal commutative quadratic refinement .
in fact , the forgetful functor @xmath226 from the category @xmath278 of commutative quadratic forms to the category @xmath220 of hermitian forms has a left adjoint @xmath279 . as hinted to above
, we will force the anti - involution @xmath161 to be trivial in the universal construction of theorem [ thm : universal ] .
this means that we should define the universal ( commutative ) group @xmath280 as the quotient of our previously used group @xmath281 by the relations @xmath282 by setting @xmath57 respectively @xmath36 to zero , these relations imply @xmath283 vice versa , these two types of equations imply the general ones and hence we see that @xmath280 is the quotient of the centrally extended group @xmath204 & m_{ee}/(m^*=m ) \ar[r ] & m_{ee}/(m^*=m ) \times_\lambda a \ar[r ] & a \ar[r ] & 1 } \ ] ] by the relations @xmath284 .
we write elements in @xmath280 as @xmath285 $ ] with the above relations understood .
it then follows that @xmath286 $ ] is a homomorphism @xmath287 ( which is in general not any more injective ) .
moreover , our original formula leads to a homomorphism @xmath288 given by @xmath289:= h_\lambda(m , a)= m+m^ * + \lambda(a , a)\ ] ] to see that this is well defined , observe @xmath290 and @xmath291 finally , we set @xmath292 $ ] to obtain a commutative quadratic refinement of @xmath6 which is proven exactly as in theorem [ thm : universal ] . to show that @xmath293 is universal , or more generally , that @xmath294 is a left adjoint of the forgetful functor @xmath169 , we proceed as in the proof of theorem [ thm : universal ] : we are given a morphism @xmath234 consisting of homomorphisms @xmath235 and @xmath236 such that @xmath295 we need to show that there is a _ unique _
homomorphism @xmath296 such that the three diagrams from the proof of theorem [ thm : universal ] commute .
we can use the same formulas as before , if we check that they vanish on our new relations in @xmath280 . for this
we ll have to use that the given quadratic group @xmath297 is _ commutative_. recall the formula @xmath298 splitting our relations into two parts as above , it suffices to show that @xmath299 the first equation follows from part ( iii ) of lemma [ lem : involutions ] and the fact that we are assuming that @xmath300 : @xmath301 for the second equation we compute : @xmath302 we used lemma [ lem : square ] for @xmath303 and hence the commutativity of @xmath139 .
the simplest case of a quadratic group is where both @xmath156 and @xmath161 are trivial .
let s call such a quadratic group @xmath175 _ symmetric_. equivalently , this means that @xmath304 ( and hence @xmath137 is commutative ) . then a quadratic form @xmath305 will automatically be _
symmetric _ in the sense that @xmath306 we call @xmath180 a _ symmetric quadratic refinement _ of @xmath6 and obtain a category of symmetric quadratic forms with a forgetful functor @xmath169 to the category of symmetric forms .
it is not hard to show that the construction in theorem [ thm : commutative universal ] gives a universal symmetric quadratic refinement @xmath293 for any given symmetric bilinear form @xmath6 .
more precisely , [ thm : symmetric universal ] any symmetric bilinear form @xmath6 has a universal symmetric quadratic refinement .
in fact , the forgetful functor @xmath226 from the category @xmath307 of symmetric quadratic forms to the category @xmath308 of symmetric forms has a left adjoint @xmath279 .
[ rem : injective ] we observe that the map @xmath309 is a monomorphism in this easiest , symmetric , case , just like it was in the hardest , non - commutative , case .
this can be seen by noting that the first set of relations @xmath310 is redundant if the involution @xmath156 is trivial .
therefore , if @xmath311 $ ] then @xmath312 must come from the second set of relations , i.e. it must be of the form @xmath313 this implies that @xmath314 and hence @xmath315 which in turn means @xmath316 .
[ ex : quadratic ] if @xmath317 then @xmath318 and @xmath319 is a canonical choice for which @xmath180 is determined by @xmath6 .
another canonical choice is @xmath320 and @xmath321 .
then a quadratic refinement of @xmath322 with this choice exists exactly for even forms , at least for free groups @xmath174 . moreover ,
if @xmath135 has no 2-torsion then a quadratic refinement is uniquely determined by the given even form . at the other extreme ,
consider @xmath323 . if @xmath174 is a finite dimensional @xmath324-vectorspace then non - singular symmetric bilinear forms @xmath6 are classified by their rank and their _ parity _ , i.e. whether they are even or odd , or equivalently , whether they admit a quadratic refinement or not . in the even case ,
quadratic forms @xmath184 are classified by rank and _ arf invariant_. this additional invariant takes values in @xmath324 and vanishes if and only if @xmath180 takes more elements to zero than to one ( thus the arf invariant is sometimes referred to as the `` democratic invariant '' ) .
if @xmath6 is odd then the following trick allows one to still define arf invariants and it motivates the introduction of @xmath203 .
let again @xmath174 be a finite dimensional @xmath324-vectorspace , @xmath325 and @xmath326 with the unique nontrivial homomorphisms @xmath327 .
then any non - singular symmetric bilinear form @xmath6 has a quadratic refinement @xmath180 and quadratic forms @xmath184 are classified by rank and an arf invariant with values in @xmath328 .
if @xmath6 is even , this agrees with the previous arf invariant via the linear inclusion @xmath329 .
consider a central group extension @xmath330 and assume that @xmath9 and @xmath174 have presentations @xmath331 respectively @xmath332 . to avoid confusion ,
we write groups multiplicatively for a while and switch back to additive notation when returning to hermitian forms .
it is well known how to get a presentation for @xmath333 : pick a section @xmath334 with @xmath335 which is not necessarily multiplicative . write a relation in @xmath174 as @xmath336 , where @xmath337 are generators of @xmath174 or their inverses , then @xmath338 where @xmath339 is a word in the generators of @xmath9 .
this equation follows from the fact that the projection @xmath340 is a homomorphism and for simplicity we have identified @xmath9 with its image in @xmath333 .
we obtain the presentation @xmath341 , \beta_\ell \ , w_\ell \rangle\ ] ] where @xmath342 and @xmath343 is the same word in the @xmath344 as @xmath345 is in the @xmath346 .
the commutators @xmath347 $ ] arise because we are assuming that the extension is central , in a more general case one would write out the action of @xmath174 on @xmath9
. it will be useful to rewrite this presentation as follows .
observe that the section @xmath348 satisfies @xmath349 for a uniquely determined _ cocycle _ @xmath350 . by induction
one shows that @xmath351 comparing this expression with the definition of the word @xmath352 in the presentation of @xmath333 , it follows that @xmath353 so that the above presentation of @xmath333 is entirely expressed in terms of the cocycle @xmath20 ( and does not depend on the section @xmath348 any more ) .
now assume that @xmath354 is a hermitian form with respect to an involution @xmath156 on @xmath9 .
then the universal ( non - commutative ) quadratic group @xmath137 from example [ ex : universal ] is a central extension as above with cocycle @xmath355 . reverting to additive notation
, we see that @xmath356 where the ordering of the summands is irrelevant because @xmath9 is central in @xmath137 . summarizing the above discussion
, we get .
[ lem : presentation ] the universal ( non - commutative ) quadratic group @xmath137 corresponding to the hermitian form @xmath6 has a presentation @xmath357 , \beta_\ell + \sum_{1\leq i < j \leq r } \lambda(a'_i , a'_j ) \rangle\ ] ] where the generators @xmath358 and words @xmath359 are defined as above .
moreover , the universal quadratic refinement @xmath360 is a ( in general non - multiplicative ) section of the central extension and hence @xmath361 for the generators @xmath346 of @xmath174 .
as discussed in theorem [ thm : commutative universal ] , we get the universal _ commutative _ quadratic group @xmath362 for @xmath6 by adding the relations @xmath363 and @xmath364 . the latter can be rewritten in the form @xmath365 . in the current notation , where @xmath312 is identified with @xmath366
, we obtain the relations @xmath367 recalling that @xmath368 and @xmath362 are _ commutative _ groups , we can write our presentation in that category to obtain [ lem : commutative presentation ] the universal ( commutative ) quadratic group @xmath280 corresponding to the hermitian form @xmath354 has a presentation @xmath369 here @xmath331 is a presentation of @xmath9 and @xmath346 are generators of @xmath174 . moreover , for every relation @xmath370 in @xmath174 , we use the word @xmath371 .
if we apply this construction to the universal inner product on order @xmath33 rooted trees @xmath372 we obtain a universal symmetric quadratic refinement @xmath373 let us compute the presentation from lemma [ lem : commutative presentation ] in this case . recall that the generators of @xmath374 are rooted trees @xmath79 of order @xmath33 and the relations are the as and ihx relations from figure [ fig : relations ] .
similarly , @xmath97 is generated by unrooted trees @xmath121 of order @xmath263 , modulo the same relations .
putting these together , we see that @xmath375 is generated by unrooted trees @xmath121 of order @xmath263 and elements @xmath376 , one for each rooted tree @xmath79 of order @xmath33 .
the three types of relations from lemma [ lem : commutative presentation ] are : * relations in @xmath377 are ordinary as and ihx relations for unrooted trees @xmath121 , * every relation @xmath345 in @xmath378 is an as - relation @xmath379 or an ihx - relation @xmath380 .
we obtain the following _ twisted _
as- respectively ihx - relations : @xmath381 * @xmath382 the last relation @xmath20 builds in the commutativity of the universal group as discussed above because we are in the easiest , symmetric , setting where the involution @xmath156 is trivial . using relation @xmath20 ,
the twisted as relation simply becomes @xmath383 which was expected since we are in the symmetric case .
this relation means that the orientation of @xmath79 is irrelevant when forming @xmath376 and in fact , with some care one can see that the twisted ihx - relation makes sense for unoriented trees .
[ lem : quadratic - form ] this is a presentation for the target group @xmath384{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } _ { 2n}(m)$ ] of twisted whitney towers from definition [ def : twisted - tree - groups ] .
the translation comes from setting @xmath106{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } = q(j)$ ] for rooted trees @xmath79 ( and keeping unrooted trees unchanged ) .
we need to show that the twisted ihx - relations in the original definition of @xmath385{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } _ { 2n}(m)$ ] are equivalent to the twisted ihx - relations above , all other relations were already shown to agree .
this is very easy to see in the presence of the interior - twist relations : together with the ( untwisted ) ihx - relations , they imply that @xmath386 this last expression is exactly the difference between the two versions of the twisted ihx - relations .
[ cor : quadratic ] there is an isomorphism of symmetric quadratic groups @xmath387{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } _ { 2n}(m)\ ] ] which is the identity on @xmath388 and takes @xmath376 to @xmath106{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } $ ] for rooted trees @xmath79 . the quadratic group structure on @xmath108{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } _ { 2n}(m)$ ] is given by the homomorphisms @xmath389{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } _ { 2n}(m)\overset{h}{\to}{\mathcal{t}}_{2n}(m)$ ] which are uniquely characterized ( for unrooted trees @xmath121 and rooted trees @xmath79 ) by @xmath390{infty2.pdf}\end{minipage } } { \begin{minipage}{.12in}\includegraphics[width=.12in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.10in}\includegraphics[width=.10in]{infty2.pdf}\end{minipage } } { \begin{minipage}{.08in}\includegraphics[width=.08in]{infty2.pdf}\end{minipage } } } } ) = \langle j , j \rangle\ ] ] | we first remind the reader of a simple geometric description of the kirby - siebenmann invariant of a 4manifold in terms of a quadratic refinement of its intersection form .
this is the first in a sequence of higher - order intersection invariants of whitney towers , studied by the authors , particularly for the 4ball . in the second part of this paper , a general theory of quadratic forms
is developed and then specialized from the non - commutative to the commutative to finally , the symmetric settings .
the intersection invariant for twisted whitney towers is shown to be the universal symmetric refinement of the framed intersection invariant . as a corollary
we obtain a short exact sequence that has been essential in the understanding of whitney towers in the 4ball . |
quantum optomechanics is the study of phenomena originating from the mutual interaction between electromagnetic radiation and mechanical motion @xcite . in cavity optomechanics both the electromagnetic field and the mechanical vibrations are effectively restricted to single modes , and the strong coupling among phonons and photons achievable there enabled the demonstration of various quantum mechanical effects during the last years @xcite . in the domain of optical frequencies strongest optomechanical coupling has been obtained in optomechanical crystals @xcite where nanostructuring of dielectric materials is exploited to generate phonon and photon modes with strong spatial localization in order to enhance the light - matter interactions .
very recently , experimental progress with nanophotonic waveguides supporting long - lived , high - frequency phonon modes evidenced that quantum optomechanical effects may become accessible also within a multi - mode version of optomechanics involving continua of propagating phonon and photon modes @xcite . at the classical level ,
the brillouin physics of interacting photons and phonons in waveguides has been studied extensively and led to the demonstration of a wide scope of nonlinear optical phenomena , see @xcite for reviews .
so far , the dominant mechanism for optomechanical coupling in waveguides has been electrostriction , that is the modulation of the index of refraction of the bulk dielectric material associated with its acoustic vibrations causing scattering of photons on these periodic index modulations .
the recent experiments with nanophotonic waveguides entered a new regime where radiation pressure effects due to vibrational surface deformations start to dominate over electrostriction which may result in vastly enhanced photon - phonon coupling . at the same time
, these devices can maintain large quality factors for ghz mechanical modes extending over long cm - scale nanowires providing large optomechanical interactions over significant time and length scales .
overall , these developments indicate brillouin quantum optomechanics as a promising route towards integrable , broad - band platforms supporting strongly interacting fields of phonons and photons . on the theoretical side , the description of brillouin optomechanics in classical terms is well established @xcite .
a corresponding quantum mechanical description of electrostrictive coupling in terms of a multi - mode hamiltonian has recently been derived by agrawal et al .
@xcite in the context of multimode phonon cooling . in cavity quantum optomechanics involving single phonon and photon modes
both effects are commonly taken into account @xcite based on a formula by johnson et al .
progress towards extending the cavity optomechanical description to the regime of brillouin optomechanics has been achieved by van laer et al
. @xcite by relating brillouin gain parameters to the optomechanical single - photon coupling strengths .
sipe et al .
@xcite very recently provided a hamiltonian treatment of stimulated brillouin scattering in nanoscale integrated waveguides accounting for electrostriction and radiation pressure , following the method of @xcite . in the present work we aim to contribute to this development of a quantum theory of brillouin optomechanics in two respects : firstly
, we derive the multi - mode hamiltonian for brillouin optomechanics by applying perturbation theory directly to the field hamiltonian for the case of an isotropic dielectric material which includes on equal footing electrostriction and radiation pressure mechanisms .
our derivation reproduces the results of sipe et al .
@xcite but avoids the rather technical smoothing procedures introduced by johnson et al .
@xcite , employed also in @xcite , in order to deal with discontinuities at surfaces .
we believe that the point of view advocated in the present derivation provides valuable physical insight to an otherwise rather unintuitive result .
secondly , we apply the general formula of the multi - mode hamiltonian for brillouin optomechanics to the important special case of a cylindrical nanowaveguide , and evaluate analytically the parameters in the hamiltonian characterizing the phonon - photon coupling strength .
we use our analytical expressions to demonstrate the domination of radiation pressure effect over electrostriction for nanoscale waveguides , and determine optimal parameter regimes exhibiting maximal coupling .
the formalism is applicable to systems of any dimensional scales , ranging from cavity optomechanical systems involving localized photon and phonon modes up to bulk materials in which photons and phonons are described by continuous fields as in brillouin optomechanics .
we treat a cylindrical nanoscale waveguide ( tapered fiber ) , which exhibits a ( quasi)continuum of modes propagating in the longitudinal direction with localized discrete modes due transverse confinement .
we consider in much detail the case of a nanofiber made of silicon material that is embedded in free space .
we analytically solve for the photon and phonon vector mode functions and dispersion relations and calculate the photon - phonon coupling parameters originating from electrostriction and radiation pressure mechanisms .
this case is motivated by the recent progress in brillouin optomechanics @xcite , but also by the work on tapered optical nanofibers that have been used in manipulating , trapping , and detecting neutral cold and ultracold atoms @xcite , in which optically active mechanical modes of tapered nanofibers have been also investigated @xcite .
moreover , we provide a means to compare the theoretically calculated photon - phonon coupling parameter with the experimentally observed gain parameter . moving to the real - space representation of the hamiltonian
, we solve the system of equations for the stokes stimulated brillouin scattering .
the stokes field is amplified with a gain parameter that is related to the photon - phonon coupling parameter in the derived hamiltonian , similar to what has been discussed by van laer et al .
@xcite .
the paper is organized as follows .
section [ sec : perturbationtheory ] contains the derivation of the classical , perturbed hamiltonians for the coupled light and mechanical excitations through radiation pressure and electrostriction mechanisms .
the discussion is followed by the canonical quantization of the coupled classical fields , where the interacting multi - mode photon and phonon hamiltonian is derived .
the photon - phonon coupling parameters are explicitly calculated for the case of cylindrical nanowires in section [ sec : nanonwire ] .
a relation between the photon - phonon coupling parameter and the gain parameter of stokes stimulated brillouin scattering is presented in section [ sec : gain ] , and the coupled photon - phonon real - space hamiltonian is introduced .
the appendices include the full solutions of the electromagnetic fields and the mechanical excitations in a cylindrical dielectric waveguide .
we aim first to derive the classical hamiltonian that represents the mutual influence of the mechanical excitations and the electromagnetic field in a bounded dielectric medium , and which are characterized by the electric field @xmath0 and the mechanical displacement field @xmath1 , respectively .
the electromagnetic field in a dielectric , lossless , and nonmagnetic medium with a scalar permittivity @xmath2 is described by the hamiltonian @xcite @xmath3 where the electric displacement field is defined by @xmath4 .
the hamiltonian coupling of photons and phonons follows from evaluating the correction @xmath5 to @xmath6 due to a mechanical displacement @xmath1 causing a perturbation @xmath7 in the permittivity @xmath2 .
we will consider corrections in first order of the mechanical displacement throughout the paper , and limit the discussion to isotropic fluctuations in the dielectric constant .
the explicit dependence of @xmath7 on @xmath1 differs for radiation pressure and electrostrictive interaction , and will be detailed below .
nevertheless , both effects are covered by the perturbation to the field hamiltonian in eq .
@xmath8 here @xmath9 in eq .
( [ deltaemh ] ) denotes the perturbation of the inverse of the permittivity which is not to be confused with the inverse of the perturbation @xmath10^{-1}$ ] .
the contributions to @xmath5 due to the corrections in the amplitude of the electric displacement field @xmath11 are negligibly small , on the order of the ratio of phonon to photon frequency , as shown in appendix [ sec : maxwell ] .
a further correction of the same order is introduced through magnetic polarization effects @xcite .
note that the perturbation of the field hamiltonian ( [ emh ] ) needs to be done in the representation where the energy density is expressed in terms of the electric displacement field .
if instead the electric field @xmath12 is used , the contributions due to corrections of the field amplitude will be significant , cf .
appendix [ sec : maxwell ] , and @xmath5 would take a much more cumbersome form .
we consider a dielectric material with permittivity @xmath13 in a volume @xmath14 , that is localized in a surrounding medium with permittivity @xmath15 occupying the complementary volume @xmath16 . for a dielectric material in vacuum @xmath17 .
the unperturbed permittivity can be written as @xmath18 where @xmath19 is a step function defined by @xmath20 a mechanical displacement @xmath1 of the dielectric medium in @xmath14 will affect the material s permittivity in two ways : in the electrostrictive mechanism , the fluctuations change the magnitude of the permittivity of the material in @xmath14 at a fixed boundary .
radiation pressure in turn corresponds to fluctuations in the material boundary at a fixed magnitude of @xmath21 .
thus , the perturbed permittivity can be written as @xmath22\theta({\bf x+{\bf q}}),\ ] ] where @xmath23 .
the first order corrections is @xmath24 and the contributions from radiation pressure and electrostriction are , respectively , @xmath25 the contributions of these two perturbations to the interaction hamiltonian in eq .
will be treated in the following two subsections .
we note first that @xmath26 in eq .
denotes a delta distribution whose effect is to turn volume integrals into surface integrals over the boundary @xmath27 between @xmath14 and @xmath16 , that is @xmath28 where @xmath29 is the infinitesimal surface element .
thus , when using eq . in it will be vital to take care that the remaining integrand does not contain discontinuities on the boundary surface rendering the integral undefined .
this concerns in particular the discontinuity in the field component of @xmath30 parallel to the boundary surface . in order to overcome this difficulty we consider an ( arbitrarily thin ) shell volume @xmath31 enclosing the boundary surface @xmath27 , and rewrite the contribution to the interaction hamiltonian within @xmath32 in terms of the fields @xmath33 and @xmath34 parallel and orthogonal to the surface that are both continuous , @xmath35 here we used @xmath36 . in restricting the integration to the shell volume @xmath32 we anticipate , in view of eq .
, that it is the energy within this subvolume which will be relevant for the perturbation due to radiation pressure .
the remaining hurdle is to arrive at a well behaved perturbation of the inverse permittivity in eq .
this can be achieved by expressing the inverse as @xmath37 which yields @xmath38 for the contribution due to radiation pressure .
finally , we can use eqs .
, , and in expression to arrive at the hamiltonian describing radiation pressure interaction @xmath39 we used the symbols @xmath40 and @xmath41 following the notation introduced by johnson et al . in @xcite .
the integral in eq .
is over the surface @xmath27 of the dielectric medium , and all fields are evaluated on that surface .
thanks to the continuity of @xmath42 and @xmath43 there is no ambiguity in evaluating the field on either side of the surface .
the result presented here agrees with the one derived in @xcite for the perturbed eigenfrequency of photonic field modes .
johnson et al . used a smoothing procedure in order to deal with the difficulty of discontinuities of field amplitudes at the surface .
the alternative approach presented here avoids such technicalities and at the same time provides directly the interaction hamiltonian covering both frequency shifts of photon modes and brillouin scattering among different field modes .
it is this last aspect which is of main interest in the description of the optomechanics of extended nanophotonic waveguides .
furthermore , the interaction hamiltonian is directly amenable to quantization , as will be done in sec .
[ sec : quantization ] , and provides firm grounds for the description of radiation pressure effects in brillouin quantum optomechanics .
electrostriction appears due to the tendency of a dielectric material to be compressed in the presence of light , and as a consequence to excite mechanical vibrations in the medium @xcite .
the appearance of mechanical vibrations modulate the optical properties and result in small changes in the dielectric constant that induce scattering of the light .
a hamiltonian description of electrostriction has been derived previously by agrawal et al .
@xcite . for completeness
we present a derivation within the present approach , starting from expression for the perturbation of the bulk value of the permittivity due to a mechanical displacement . by means of the electrostriction
constant @xmath44 the change in the permittivity can be related to fluctuations of the mass density @xmath45 @xmath46 which in turn is determined through the mechanical displacement @xmath47 overall , we arrive from eq .
( [ eq : deltaepsel ] ) at @xmath48 electrostriction mechanism in nanoscale structures can give rise to anisotropic phenomena .
anisotropic contributions induced by longitudinal phonons in nanoscale waveguides are immaterial @xcite .
as we concentrate mainly in processes involving longitudinal phonons we treat here only the isotropic case of scalar fluctuations @xcite .
anisotropic phenomena induced by tensor fluctuations of the dielectric function are beyond the scope of the present paper .
the relation ( [ eq : deltaepsel ] ) can be used directly in eq .
( [ deltaemh ] ) when resorting to the representation of @xmath5 in terms of the electric field ( using again @xmath49 ) @xmath50 no issues regarding discontinuities in the integrand arise here since the domain of integration is effectively limited to the volume @xmath14 occupied by the dielectric due to the step function in eq .
( [ eq : deltaepsel ] ) .
here @xmath51 decreases by increasing @xmath52 as expected .
finally , this yields the hamiltonian for the electrostrictive interaction of photons and phonons , @xmath53 it is evident from eqs .
( [ hamrp ] ) and ( [ hamel ] ) that radiation pressure and electrostriction are surface and volume effects , respectively . for sufficiently small dimensions
radiation pressure will therfore dominate , and may ultimately provide largely enhanced coupling strengths per single photon and phonon .
we will show this explicitly for the example of a cylindrical waveguide in sec .
[ sec : nanonwire ] . before that , we will quantized the interaction hamiltonians in eqs .
( [ hamrp ] ) and ( [ hamel ] ) , and extract the quantum mechanical coupling strengths at the single photon / phonon level .
we aim to derive the photon - phonon interaction hamiltonian in dielectric media by canonically quantizing the electromagnetic and mechanical fields that were presented in the previous sections .
the hamiltonian of the free photon field reads @xcite @xmath54 where the summation is over all the photon modes and the summation index @xmath55 comprises any index labeling photon modes for a given geometry . in the following
we will assume a discrete set of indices , such that @xmath56 and @xmath57 are dimensionless bosonic creation and annihilation operators , and fulfill @xmath58=\delta_{\alpha\beta}$ ] .
continuous index sets as relevant to problems with _
e.g. _ translational symmetry can be attained via appropriate limiting procedures , as detailed for example in ref .
the electric field operator reads @xmath59 , where @xmath60 and @xmath61 .
the ( dimensionless ) vector function @xmath62 of mode @xmath55 and the eigenfrequency @xmath63 are obtained by solving maxwell s equations with appropriate boundary conditions , cf .
appendix [ sec : maxwell ] .
the electric field amplitude of a single photon in mode @xmath55 is @xmath64 the effective volume @xmath65 is calculated from the normalization condition @xmath66 on the other hand , the phonon hamiltonian reads @xmath67 where @xmath68 and @xmath69 are the bosonic creation and annihilation operators of a phonon in mode @xmath70 with angular frequency @xmath71 .
the index @xmath70 again stands for any discrete index relevant for a given geometry .
the operator corresponding to the mechanical displacement field is @xmath72 , where @xmath73 and @xmath74 .
here @xmath75 is the ( dimensionless ) vector mode function of phonons with eigenfrequency @xmath71 which is obtained by solving the equations of motion of mechanical excitation as shown in sec .
[ sec : nanonwire ] .
the zero - point - fluctuation of mode @xmath70 is @xmath76 the effective mass associated to phonon mode @xmath70 is given by @xmath77 , where @xmath45 is the medium s mass density and @xmath78 is the phonon effective volume , which can be calculated from the normalization condition @xmath79 the quantized interaction hamiltonian is obtained from the classical one by replacing the displacement and the electromagnetic fields by operators , and using normal ordering .
we will use the notation @xmath80 to denote the normally ordered form of the operator @xmath81 .
the radiation pressure hamiltonian reads @xmath82 and the electrostriction hamiltonian reads @xmath83 in terms of creation and annihilation operators , using the above definitions , the interaction hamiltonian is @xmath84 thus , the optomechanical photon - phonon coupling strength ( of dimension hz ) is @xmath85 , where the radiation pressure coupling is given by @xmath86 all mode functions are evaluated on the boundary surface , and most important here is that the perpendicular components , @xmath87 , are evaluated on the internal side of the surface , as indicated by the superscript symbol @xmath88 . for the electrostrictive coupling
we get @xmath89 the results are applicable to any dielectric material ranging from fully confined media , as for a resonator , up to partly confined media , as for a waveguide .
the interactions are consistent with the results in @xcite , but here we give explicitly the appropriate normalized amplitudes .
the formalism of the previous section has been applied extremely successfully to quantum optomechanical systems comprising single electromagnetic and mechanical modes .
our main concern here is the application to extended media , where the multimode character of the interaction hamiltonian is essential but at the same time field confinement due to small spatial extensions yields strong coupling due to enhanced radiation pressure . in the following we present analytical results for the coupling strengths @xmath90 and @xmath91 for the most elementary such geometry , namely a cylindrical nanophotonic waveguide .
we consider a dielectric nanowire that is localized in free space and extended along the @xmath92 direction over a length @xmath93 with a nanoscale radius @xmath94 , as seen in figure ( [ fiber ] ) .
the dielectric constants are @xmath95 and @xmath17 , where @xmath96 is the medium refractive index .
recently , such tapered optical nanofibers have been intensively studied @xcite . the strong confinement in the transverse direction results in discrete modes for both the electromagnetic and mechanical fields . in the following we will consider only a single transverse mode for the photons and phonons .
the electromagnetic and mechanical fields can propagate along the waveguide axis with wavenumber @xmath97 which takes on the values @xmath98 with @xmath99 , where @xmath93 is the waveguide length and we use periodic boundary conditions .
phonon wavenumbers will be denoted by @xmath100 , and have a natural cut - off that is given by the inverse of the crystal lattice constant . and refractive index @xmath96 , which is localized in free space . ]
the photon hamiltonian for a single tranverse mode reads @xmath101 where @xmath102 and @xmath103 are the creation and annihilation operators of a photon of wavenumber @xmath97 and angular frequency @xmath104 .
the electric field operator reads , in cylindrical coordinates , @xmath105 where @xmath106 is the transverse vector mode function , and the effective mode volume @xmath107 is defined by @xmath108 .
the lowest propagating mode in an optical nanofiber is the he@xmath109 mode .
we concentrate in photons with a fixed polarization , and we consider rotating polarization with left or right hand circular rotations .
detailed calculations of the cylindrical waveguide photon dispersions and mode functions are given in appendix [ app : fibrefield ] .
the photon dispersion , which is the relation between the angular frequency , @xmath110 , and the wavenumber along the fiber axis , @xmath97 , can be extracted from the expression @xcite @xmath111 where @xmath112 , and @xmath113 , with @xmath114 .
propagating modes can appear only in the range @xmath115 . in the literature
this result is commonly represented in terms of the fundamental parameter @xmath116 . for silicon
we have @xmath117 and up to about @xmath118 only the he@xmath109 photons propagate in the fiber , and beyond @xmath118 tm and te modes can be excited .
the he@xmath109 dispersion relation is plotted in figure [ photdis ] for @xmath119 as a function of @xmath120 . for small wavenumbers the photons
are unconfined in the nanofiber and propagate with the group velocity @xmath121 , but beyond @xmath122 they get confined and propagate with almost linear dispersion of group velocity @xmath123 .
fiber mode is presented as @xmath119 vs. @xmath120 . here
@xmath124 for silicon material . beyond the dashed line at @xmath125
hz other photon branches start to appear . ]
the vector mode functions inside the fiber , that is for @xmath126 , are given by @xmath127e^{\pm i\theta } , \nonumber \\ u^{\theta<}_k&=\pm b \frac{k}{2p}\left[(1-s)j_0(pr)+(1+s)j_2(pr)\right]e^{\pm i\theta},\nonumber \\ u^{z<}_k&=bj_1(pr)e^{\pm i\theta},\end{aligned}\ ] ] and outside the fiber , that is @xmath128 , are given by @xmath129e^{\pm i\theta } , \nonumber \\ u^{\theta>}_k&=\pm b \frac{k}{2q}\frac{j_1(pa)}{k_1(qa ) } \nonumber \\ & \left[(1-s)k_0(pr)-(1+s)k_2(qr)\right]e^{\pm i\theta } , \nonumber \\
u^{z>}_k&=b\frac{j_1(pa)}{k_1(qa)}k_1(qr)e^{\pm i\theta},\end{aligned}\ ] ] where @xmath130\left\{\frac{j_1^{\prime}(pa)}{paj_1(pa)}+\frac{k_1^{\prime}(qa)}{qak_1(qa)}\right\}^{-1},\ ] ] and @xmath131 stand for left and right hand circular polarizations .
the parameter @xmath132 is fixed from the normalization relation stated above . in figure [ volphot ]
we plot the effective volume @xmath107 of photon mode @xmath97 relative to the total fiber volume @xmath133 as a function of @xmath120 . here
a minimum appears around @xmath134 in which the photon mode is highly concentrated inside the fiber , and a small part penetrates outside .
for example , for @xmath135 nm we get a minimum at @xmath136 nm .
vs. @xmath120 for he@xmath109 fiber photons . ]
the phonon hamiltonian for a single mode is given by @xmath137 where @xmath138 and @xmath139 are the creation and annihilation operators of a phonon of wavenumber @xmath100 and angular frequency @xmath140 .
the displacement operator is defined by @xmath141 where @xmath142 is the transverse vector mode function .
the zero - point - fluctuation is @xmath143 with effective mass @xmath144 and effective phonon mode volume @xmath145 . in optical nanofibers
torsional , longitudinal and flexural phonons can be excited .
here we consider only longitudinal modes , as the torsional modes decouple to the light through radiation pressure and the flexural modes are of higher energy .
detailed calculations of the cylindrical waveguide phonon dispersions and mode functions are given in appendix [ app : fibrephonons ] .
the longitudinal phonon dispersion can be extracted from the expression @xcite @xmath146 where we have @xmath147 , and @xmath148 .
the lowest two longitudinal branches are plotted in figure [ phondis ] for @xmath149 as a function of @xmath150 .
we treat silicon material with @xmath151 m / s and @xmath152 m / s . for small wavenumbers @xmath153 ,
the lowest acoustic modes have a linear dispersion , and the lowest vibrational modes are almost dispersion - less up to @xmath154 .
both branches become linear beyond the anti - crossing point . vs. @xmath150 for a silicon waveguide ( @xmath124 ) :
lower , acoustic branch ( blue ) and upper vibrational branch ( red ) . ]
the vector mode functions are given by @xmath155 the parameters @xmath156 and @xmath157 are fixed using the boundary condition relation @xmath158}\frac{j_1(\eta_la)}{j_1(\eta_ta)}a,\ ] ] and the normalization relation stated above . note that @xmath156 also plays the role of a scaling factor that takes care of the appropriate units .
vs. @xmath150 , ( a ) for acoustic modes , and ( b ) for vibrational modes.,title="fig : " ] vs. @xmath150 , ( a ) for acoustic modes , and ( b ) for vibrational modes.,title="fig : " ] for illustration we consider a nanofiber of radius @xmath135 nm and length @xmath159 cm made of silicon material ( density @xmath160 kg / m@xmath161 ) . in figures [ zpf1].a and [ zpf1].b we plot the zero - point - fluctuation relative to the fiber radius , @xmath162 , as a function of @xmath150 , for the lowest two branches .
it appears that the zero - point fluctuations decrease with increasing @xmath150 for the acoustic modes , where a singularity appears around @xmath163 , and increase afterward for larger @xmath150 .
the singularity appears at the anti - crossing point among the acoustic and the lowest vibrational modes , as seen in figure [ phondis ] .
the vibrational modes have finite zero - point - fluctuations that decrease with increasing @xmath150 .
importantly , the phonon frequencies for both acoustic and vibrational modes are in the ghz regime which allows to achieve low thermal occupation numbers at cryogenic temperatures . moreover , mechanical quality factors measured in recent experiments with tapered nanofibres were in the range of @xmath164 @xcite . for square - shaped waveguides mechanical quality factors of several 100 have been measured @xcite . in view of the tremendous progress made regarding mechanical quality factors of single - mode optomechanical systems
we expect that there is vast room for improvement here with optimized nanostructures . inserting the analytical expressions for the nanowire photon and phonon modes derived in the previous sections into the general expression for the photon - phonon interaction hamiltonian yields @xmath165 where we exploited the translational symmetry along the waveguide axis implying @xmath166 translational symmetry results in conservation of momentum in which two photons of wavenumbers @xmath97 and @xmath167 scatter by emission or absorption of a phonon of wavenumber @xmath100 .
contour vs. the plane @xmath168 , for scattering that involves : ( a ) acoustic modes , and ( b ) vibrational modes .
the fiber length is @xmath159 cm .
the optomechanical parameters @xmath169 scale as @xmath170 .
, title="fig : " ] contour vs. the plane @xmath168 , for scattering that involves : ( a ) acoustic modes , and ( b ) vibrational modes .
the fiber length is @xmath159 cm .
the optomechanical parameters @xmath169 scale as @xmath170 .
, title="fig : " ] the coupling is @xmath171 , as in eq . .
the coupling parameter due to radiation pressure is given by @xmath172 where @xmath173 and all vector mode functions are evaluated on the fiber surface according to eqs . .
the effective radius of the photon mode @xmath174 is defined through @xmath175 . in view of @xmath176 we approximate @xmath177 in the following .
the dependence of @xmath178 on @xmath93 is only through @xmath179 that gives @xmath180 . in figs .
[ coup1].a and [ coup1].b we show the radiation pressure coupling parameter , @xmath181 , versus @xmath120 and @xmath150 for scattering that involves , respectively , the acoustic modes and the lowest vibrational modes for a fiber of length @xmath159 cm .
radiation pressure coupling parameters have high values of about @xmath182 khz , in the region of @xmath183 , for acoustic modes and @xmath184 for vibrational modes , which appear in the region of @xmath185 for photons .
this region for optical photons appears at nanoscale waveguides , and the coupling is significantly decreased at microscale and larger structures .
the coupling parameter due to electrostriction is given by @xmath186 where @xmath187 in figs .
[ coupel1]a and [ coupel1]b we show the coupling parameter @xmath188 versus @xmath120 and @xmath150 , for scattering involving , respectively , the acoustic modes and the lowest vibrational modes again for a fiber of length @xmath159 cm .
the electrostriction parameter for dielectric materials can be written as @xmath189 , where @xmath190 is the elasto - optic parameter . for silicon
we have @xmath117 and @xmath191 , hence we get @xmath192 . for acoustic modes
the electrostriction coupling parameters increase with increasing @xmath120 and @xmath150 , while for vibrational modes they increase with increasing @xmath120 only at small @xmath150 .
the comparison between the two coupling mechanisms show that electrostriction is small in zones where radiation pressure is maximal , which is the case for optical light in nanoscale waveguides .
but for microscale and larger waveguides electrostriction becomes dominant where radiation pressure is significantly suppressed .
nanowires of square cross sections are expected to give larger photon - phonon coupling parameters due to the fact that light is more concentrated on the boundary @xcite , but they have smaller mechanical quality factor relative to cylindrical nanofibers .
contour vs. the plane @xmath168 , for the scattering that involves : ( a ) acoustic modes , and ( b ) vibrational modes .
the fiber length is @xmath159 cm .
the optomechanical parameters @xmath193 scale as @xmath170,title="fig : " ] contour vs. the plane @xmath168 , for the scattering that involves : ( a ) acoustic modes , and ( b ) vibrational modes .
the fiber length is @xmath159 cm .
the optomechanical parameters @xmath193 scale as @xmath170,title="fig : " ]
in this section we transform the coupled photon - phonon hamiltonian from momentum - space to real - space representation .
this is especially instructive in cases when the description can be effectively constrained to relatively narrow frequency bands , as is the case when narrowband light is injected into the nanofibre and at the same time brillouin scattering populated only selected narrow bands of phonons . the real - space representation developed in the following provides coupled one dimensional propagation equations for narrowband photons and phonons . as a first application we derive the brillouin gain parameters for nanofibres measured in recent experiments from our ab initio calculation of the optomechanical coupling parameters @xmath169 and @xmath193 .
for the effectively one - dimensional photon field introduced in sec .
[ sec : phot ] we define associated operators in real space as @xmath194 here @xmath195 denotes a suitable bandwidth of photon wave numbers centered around a central wave numbers @xmath196 .
the operator @xmath197 defined such as to describe slowly varying spatial amplitudes relative to the wave @xmath198 . for positive ( negative ) sign of @xmath196 the slowly varying operators @xmath197
describe right ( left ) propagating photons .
the definition of @xmath197 implies @xmath199=i\delta(z - z')$ ] where the @xmath200-function is understood to be of width @xmath201 .
the inverse relation is @xmath202 furthermore , we approximate the photon dispersion relation shown in fig .
[ photdis ] within the relevant bandwidth @xmath195 as @xmath203 , where @xmath204 is the bandwidth s central frequency , and @xmath205 is the group velocity . for the silicon nanowire considered here we have @xmath206 in the range of wave numbers @xmath207 , cf .
[ photdis ] . in real space representation
the hamiltonian of the free photon field for modes within the bandwidth @xmath195 is @xmath208 in complete analogy , we define for the one dimensional fields in sec .
[ sec : phon ] of acoustic phonons and vibrational phonons the real space operators @xmath209 with inverse relation @xmath210 and commutation relation @xmath211=\delta(z - z')$ ] .
the phonons have a linear dispersion with sound velocity @xmath212 , such that within the bandwidth @xmath213 we approximate @xmath214 .
thus , the free hamiltonian for phonons is @xmath215 the vibrational modes are almost dispersion - less below @xmath216 such that @xmath217 in this regime .
finally , the interaction hamiltonian is given by @xmath218 where @xmath219 is the photon - phonon coupling parameter in the local field approximation , in which we neglect the weak dependence of @xmath219 on wavenumbers within the relevant bandwidths @xmath213 and @xmath195 of phonon and photon modes . and
wave number @xmath220 is scattered into the stokes field of frequency @xmath221 and wave number @xmath222 , and a sound wave of frequency @xmath223 and wave number @xmath100 . ] the real space description from the previous section generalizes immediately when more than one band of phonon or photon modes are considered , as we will discuss now for the case of stimulated brillouin scattering .
the result of this treatment will be a direct relation between the optomechanical coupling parameters @xmath224 and the brillouin gain parameter which can be observed directly in experiments .
we consider the backward brillouin scattering among two narrow band light fields , @xmath225 the stokes probe field and @xmath226 the strong pump field , involving acoustic phonons , as seen in fig .
we denote the central frequencies for the three bands by @xmath221 , @xmath227 and @xmath223 , respectively , and assume energy conservation , @xmath228 , and momentum conservation , @xmath229 .
these conditions are illustrated in fig .
[ pumpstokes ] .
generalizing the results from the previous section to this configuration , the real - space hamiltonian is @xmath230 this hamiltonian is written in an interaction picture with respect to @xmath231 and under a rotating wave approximation where all non - resonant terms ( such as @xmath232 or @xmath233 ) were dropped .
the equations of motion corresponding to the hamiltonian in eq .
are @xmath234 where @xmath235 is the acoustic phonon damping rate , and we neglect the photon damping .
@xmath236 is the langevin noise operator .
the damping rate parameter can be extracted from the observed mechanical quality factor @xmath237 and in knowing the phonon frequency @xcite . at steady state
all time derivatives for the slowly varying operators vanish and , moreover , @xmath238 can be neglected for acoustic phonons .
hence we get @xmath239 substitutions of @xmath240 in the equations of motion for the photon fields yields @xmath241 the light field intensity is defined by @xcite @xmath242 where we set @xmath243 , and denote by @xmath244 the waveguide cross section .
the strong pump field is considered as a classical field , and we use the langevin force properties @xmath245
. then we have @xmath246 where the brillouin gain factor is defined by @xmath247 neglecting the pump depletion , then the first of eqs .
can be integrated to give @xmath248 .
note that the stokes field is propagating to the left , such that @xmath249 described the outgoing intensity .
thus , the gain parameter @xmath250 expresses the gain in stokes intensity per medium volume and pump intensity .
we remark also that the coupling parameters in eqs . ) and are proportional to @xmath170 which makes @xmath251 and therefore also @xmath250 independent of @xmath93 . and
momentum @xmath252 . ]
we consider the scattering of the photons maximally localized in transverse direction with @xmath134 that is , for @xmath253{nm}$ ] , of wave length @xmath254 nm . the phonons for the backward scattering are of @xmath255 with the damping rate of @xmath256{mhz}$ ] .
the photon - phonon coupling parameter is @xmath257{khz}$ ] , for a nano - waveguide of length @xmath258{cm}$ ] . for @xmath259{ghz}$ ]
the quality factor is @xmath260 .
the group velocity is about @xmath123 .
the gain efficiency is about @xmath261{m^{-1}w^{-1}}$ ] , which agrees with the experimental results @xcite .
the values of all parameters used here are summarized in table .
the result provides a relation between the observable gain parameter @xmath250 and the photon - phonon coupling parameter @xmath219 , which agree with the one derived in @xcite .
the relations allow us to compare the value of the calculated photon - phonon coupling parameter to the experimental value of the gain parameter .
a similar result holds for the forward brillouin scattering involving vibrational modes .
.values for case study on silicon nanowire [ cols="^,^,^ " , ] [ table ]
starting from the classical electromagnetic field hamiltonian in dielectric media we derived the mutual coupling between light and mechanical excitations .
the interaction hamiltonian is obtained by perturbing the medium due to mechanical excitations .
we treated two type of fluctuations in the dielectric medium .
the first is the fluctuation in the dielectric constant that generates electrostriction coupling , and the second is fluctuations in the dielectric material boundaries that introduces radiation pressure . in order to overcome the difficulty due to the field jumps on the boundaries we expressed the hamiltonian in terms of continuous fields .
the main objective in both cases is to represent the fluctuations in terms of the mechanical displacement vector .
the derived hamiltonian can be adopted for any structure of dielectric medium ranging from nano resonators up to bulk materials . in quantum mechanics the electromagnetic field
is represented as photons and the mechanical excitations as phonons , where the classical formulation allows direct quantization by converting the classical electric field and the displacement field into operators .
we treated in much details the case of a cylindrical nanophotonic waveguide , in which photons and phonons can propagate along the waveguide with continuum wavenumbers , and are strongly confined in the transverse direction with discrete modes .
we explicitly solved for the photon and phonon dispersions and mode functions , where each discrete mode gives rise to photon or phonon branch .
we derived the multi - mode photon - phonon interaction hamiltonian for both electrostriction and radiation pressure couplings .
the lowest photon branch is for he@xmath109 photons , in which a single photon mode can propagate within the fiber with linear dispersion of group velocity @xmath262 , where for a silicon nanowire this mode is maximally localized inside the waveguide around @xmath134 with a small part that penetrates into the surrounding environment .
the lowest phonon branch is of acoustic modes with linear dispersion , and the lowest excited phonon branch is dispersion - less of localized vibrational modes with frequency of about @xmath263{ghz}$ ] .
these two phonon branches approach different linear dispersions beyond the anti - crossing point .
we calculated the photon - phonon coupling parameters due to electrostriction and radiation pressure mechanisms for a silicon nanofiber .
for optical light we found that radiation pressure dominates electrostriction coupling at nanoscale waveguides with coupling parameter of about @xmath263{khz}$ ] , while electrostriction becomes dominant at larger dimensions .
we provide a tool for checking the validity of the theoretically predicted coupling parameters by relating them to the experimentally observed gain parameter .
we consider stokes backward brillouin scattering , where a strong pump field scatters into a stokes field and a sound wave .
starting from the real space representation of the photon and phonon hamiltonians , we solve the field equations of motion at steady state to get the stokes field amplification where the gain parameter is expressed in terms of the photon - phonon coupling parameter .
the results presented here provide a general outline for deriving quantum optical multi - mode hamiltonians for interacting photons and phonons in optomechanical nanophotonic structures . in view of the success achieved in recent years with optomechanical structures involving single phonon and photon modes we envision that their multimode counterparts offer great possibilities to observe and exploit quantum effects in extended nanophotonic media .
in particular , the optomechanical kerr - nonlinearity mediated by phonons can be exploited for the study of quantum nonlinear optics and for many - body physics of strongly correlated photons .
moreover , brillouin induced transparency with the possibility of slow light , in analogy to eit in cold dense atomic gases , can be achieved in the present nanoscale waveguides .
the quantum optical hamiltonian derived in the present work provides firm grounds for future explorations into these directions .
we thank raphael van laer for useful comments .
this work was funded by the european commission ( fp7-programme ) through iquoems ( grant agreement no
. 323924 ) .
we acknowledge support by dfg through quest .
we will show here that eq .
provides the relevant part of the correction to the field hamiltonian for a given change of the permittivity , and determine the order of magnitude of the correction due to induced changes in the field amplitudes .
the equations of motion following from the field hamiltonian are maxwell s equations @xmath264 where @xmath12 and @xmath265 are the electric and magnetic fields , and @xmath2 is the ( dimensionless ) permittivity of the assumed dielectric , lossless , and nonmagnetic material .
maxwell s equations imply for the magnetic field @xmath266 harmonic solutions @xmath267 therefore fulfill the eigenvalue equation @xmath268 the operator on the left hand side is hermitian @xcite and , therefore , the eigenmodes are orthonormal , @xmath269 when the permittivity is changed slightly @xmath270 the perturbed eigenmodes @xmath271 will remain orthonormal which implies for the first order corrections @xmath272 on the other hand , the electric displacement field @xmath273 corresponding to field mode @xmath274 is @xmath275 due to eq . , and the same relation holds for the first order perturbation @xmath276 . note that this statement does not apply in the same way to the electric field @xmath277 whose perturbation contains , apart from @xmath278 , further contributions proportional to @xmath7
for a general electric displacement field @xmath279 with amplitudes @xmath280 one therefore finds @xmath281 where we used the eigenvalue equation in the last step .
overall , this implies @xmath282 where we used eq . and the fact that the optical photon frequency is not changed appreciably in brillouin scattering , that is @xmath283 .
thus , the contribution to the perturbation of the field hamiltonian due to the corrections in the field amplitudes , @xmath284 , is at most of magnitude @xmath285 to first order in @xmath52 ( or equivalently , in the mechanical displacement @xmath286 ) .
note that this is not the case for @xmath287 for the reason given above .
the reasoning presented here also shows that the first order correction of the hamiltonian due to the perturbation of the magnetic field amplitudes vanishes exactly due to .
here we present the rigorous derivation of the nanofiber modes @xcite for a cylindrical waveguide .
we start from the maxwell s equations in cylindrical coordinates @xmath288 .
the fields @xmath289 , @xmath290 , @xmath291 and @xmath292 can be expressed in terms of the axial fields @xmath293 and @xmath294 by @xmath295 where @xmath296 is the axial wavenumber .
the solution for the axial fields is given by @xmath297 where @xmath298 .
the radial function @xmath299 obeys the bessel equation @xmath300 where @xmath301 .
the solution inside the fiber , that is @xmath302 , with @xmath303 , is given by @xmath304 where @xmath305 , with @xmath114 is the free space wavenumber .
the solution on the outside of the fiber , that is @xmath128 , with @xmath306 , is given by @xmath307 where @xmath308 .
using the relations ( [ eztr ] ) , the solutions inside the fiber , that is @xmath302 , are given by @xmath309 and @xmath310 the solutions outside of the fiber , that is @xmath311 , are given by @xmath312 and @xmath313 the plus sign @xmath314 is refers to the right - handed solution of the transverse field , and the minus sign @xmath315 refers to the left - handed solution .
the linearly polarized solution can be composed as a superposition of right and left - handed solutions .
the constants @xmath156 , @xmath132 , @xmath157 and @xmath316 can be fixed by the boundary condition between the inside and outside fields on the fiber surface , where the fields @xmath293 , @xmath289 , @xmath294 and @xmath290 are continuous on the boundary .
for the constants we obtain the relations @xmath317\nonumber \\ & \times\left(\frac{j_l^{\prime}(pa)}{paj_l(pa)}+\frac{k_l^{\prime}(qa)}{qak_l(qa)}\right)^{-1},\end{aligned}\ ] ] and @xmath156 can be fixed by the normalization condition . the continuity leads to a characteristic equation that determines the wavenumber @xmath296 .
for the eh modes , in which @xmath293 is larger than @xmath294 , we get @xmath318 and for the he modes , in which @xmath293 is smaller than @xmath294 , we get @xmath319 the equations can be solved numerically or graphically and give a set of discrete solutions for @xmath296 at each @xmath320 that are specified by index @xmath321 .
the modes are labeled by he@xmath322 and eh@xmath322 .
the transverse modes eh@xmath323 are usually denoted by tm@xmath323 , where @xmath294 vanishes .
the transverse modes he@xmath323 are usually denoted by te@xmath323 , where @xmath293 vanishes .
the modes he@xmath322 and eh@xmath322 are termed hybrid modes , as all field components are non - zero .
the hybrid modes represent screw rays and @xmath320 is associated with the orbital angular momentum along the fiber axis .
we present the elastic waves in a cylindrical waveguide of isotropic medium @xcite .
the displacement vector is given by @xmath324 .
in the theory of elasticity the displacement can be derived directly from the scalar @xmath325 and vector @xmath326 potentials . in general
we can make the decomposition @xmath327 , where @xmath328 is a divergence - free vector , and @xmath329 is an irrotational vector .
hence , the material displacements are obtained from the potentials by @xmath330 the scalar field obeys the wave equation @xmath331 where @xmath332 is the velocity of the longitudinal wave , in which @xmath333 .
the vector potential obeys the wave equation @xmath334 where @xmath335 is the velocity of the transverse wave , in which @xmath336 .
as the three displacements are expressed in terms of four scalar potentials , one need an extra relation between the potentials .
the constrain condition usually used is @xmath337 , but the @xmath338 condition is also possible .
the two displacement parts propagate independently . the longitudinal component , @xmath339 , propagates with velocity @xmath332 , and the transverse component , @xmath340 , propagates with velocity @xmath335 .
now we concentrate in the solution of the wave equations in cylindrical coordinates . for harmonic waves of frequency @xmath223
, we have @xmath341 with @xmath342 .
the laplacian of scalar and vector potentials are defined by @xmath343 the wave equations are given explicitly by @xmath344 the equations for @xmath325 and @xmath345 are separated , and for @xmath346 and @xmath347 are coupled .
the laplacian @xmath348 is given in cylindrical coordinates .
the solution for @xmath325 has the form @xmath349 where @xmath299 and @xmath350 obey the equations @xmath351 with @xmath352 .
the general solution has the form @xmath353 with @xmath147 .
the solution for @xmath354 can be either @xmath355 or @xmath356 , and for @xmath299 the solution is the bessel functions of the first kind .
similar solution holds for @xmath345 , where @xmath357 with @xmath148 .
we have four constants @xmath156 , @xmath132 , @xmath157 and @xmath316 . we still free to add another relation between the potentials . but
the usually used relation of @xmath337 gives a complex relation .
here we use the simple relation @xmath360 that lead to @xmath361 .
the displacement components are given in terms of the potentials by @xmath363 the boundary condition implies the surface , at @xmath364 , to be free , which leads to the equation that relates @xmath223 , @xmath100 and @xmath96 .
three cases can be treated , which are ( i ) torsional waves for @xmath365 where @xmath366 is independent of @xmath367 , ( ii ) longitudinal waves for @xmath365 where @xmath368 and @xmath369 are independent of @xmath367 , and ( iii ) flexural waves for @xmath370 where @xmath368 , @xmath366 and @xmath369 are dependent on @xmath288 . for axially symmetric torsional waves with @xmath365 , which are the simplest case for elastic waves in a rod ,
the displacement along the @xmath367 direction is defined by @xmath371 and @xmath372 , where the potentials @xmath325 , @xmath346 and @xmath347 are taken to be zero . here
the solution for the potential is @xmath373 , and has the form @xmath374 where @xmath132 is an amplitude , and @xmath375 .
the displacement reads @xmath376 the force - free surface leads to the condition @xcite @xmath377 , where either @xmath378 or @xmath379 .
the case of @xmath378 gives the solution @xmath380 with @xmath381 .
hence , we have @xmath382 the condition @xmath383 gives a set of roots , e.g. , @xmath384 , @xmath385 , @xmath386 , @xmath387 , and we write @xmath388 , where we can write the solutions as @xmath389 with @xmath390 . for longitudinal acoustic modes with @xmath365 from symmetry consideration we have @xmath391 ,
@xmath392 and @xmath393 .
here we have @xmath394 , and hence the displacements are related to the potentials by @xmath395 the solutions for the potentials are @xmath396 where @xmath397 , for @xmath398 .
the requirement of the stress to vanish on the fiber boundary , @xmath399 , lead to the pochhammer frequency equation @xmath400 in the limit of @xmath153 one gets the sound wave linear dispersion @xmath401 .
the displacements are given explicitly by @xmath402 from boundary condition we can get also relations between @xmath156 and @xmath157 . namely , the vanishing of the stress and the strain on the surface yields @xcite @xmath403j_1(\eta_ta)c.\ ] ] | we develop a systematic method for deriving a quantum optical multi - mode hamiltonian for the interaction of photons and phonons in nanophotonic dielectric materials by applying perturbation theory to the electromagnetic hamiltonian .
the hamiltonian covers radiation pressure and electrostrictive interactions on equal footing . as a paradigmatic example
, we apply our method to a cylindrical nanoscale waveguide , and derive a hamiltonian description of brillouin quantum optomechanics .
we show analytically that in nanoscale waveguides radiation pressure dominates over electrostriction , in agreement with recent experiments .
the calculated photon - phonon coupling parameters are used to infer gain parameters of stokes brillouin scattering in good agreement with experimental observations . |
for more than 35 years the spectroscopic research of bulgarian astronomers has been mainly based on data obtained with the high - quality coude spectrograph , specifically designed for the 2 meter telescope of the national astronomical observatory rozhen ( nao ) .
description of the basic characteristics of the coude spectrograph can be found in kolev ( 2008 ) and at the nao webpage .
the spectrograph was designed in the 1970s when the common standard detectors were the photoplates .
based on the data obtained with the coude spectrograph , a great number of original results have been published in different fields of research , during the era of photoplates .
it is not possible to mention all of them , but it is worthed to cite at least several papers , mainly in order to illustrate the process of transition from the era of photoplates to ccds .
properties of the stellar wind have been analysed by markova ( 1986 ) , tomov et al .
( 1990 ) explained the strange behaviour of the unique star mwc 560 , peculiar profiles of hydrogen lines were found in the spectra of several @xmath1 bootis stars , investigated by iliev and barzova ( 1993 ) , stateva ( 1995 ) explained the helium distribution on the surface of the cp4 star hd 21699 . immediately after changing from photoplates to ccds , it was clear that we need an echelle spectrograph to continue doing good science with the 2-m telescope .
markov ( 2011 ) made a summary of the scientific motivation and a review of possible solutions for an echelle spectrograph for the 2-meter telescope of nao .
the ccds were introduced in the coude spectrograph in 1993 and today an 1kx1k photometrics camera is used .
the ccd highly improved the spectrograph light efficiency and the spectral quality ( signal - to - noise ratio ) at the cost of the reduced spectral range available with a single exposure .
originally designed to cover a spectral window of 1000 and 1500 on a photographic plate ( using respective gratings ) the ccd image nowadays covers spectral ranges of 200 and 100 with a resolving power ( r ) of 15000 and 30000 respectively . evidently , with a medium - sized ccd detector the coude spectrograph offers about an order of magnitude smaller spectral range .
but most of the science cases require acquisition of the entire spectral range , simultaneously , from the uv to the near ir.one more disadvantage of coude spectrographs concerning their design is that these instruments are hosted in a large room which ensures large inter - components distances needed to achieve high spectral resolution . the large room control of working parameters ( temperature , humidity etc
is a big challenge .
a relevant solution for high - resolution spectral investigations challenges are fiber - fed echelle spectrographs .
the efficiency of these instruments is twofold .
first , their 2-d spectrum image is well suited to use effectively the full area of a 2-d ccd , thus offering the potential for the entire visible spectral range to be covered with a single exposure . and
second , their design highly reduces the hosting space , thus improving spectrograph environment control .
the first design of an echelle spectrograph for the 2-m telescope was a white pupil instrument , similar to feros ( kaufer et al .
, 1999 ) or to elodie ( baranne et al .
due to strong reduction of the preliminary planned budget for the spectrograph , we were pressed to develop a new design , different from the initial one , and based on commercial , off - the - shelf products , for all components where this was possible . finally , our echelle spectrograph got a design similar to musicos ( baudrand and bohm , 1992 ) or to the university of poznan spectrograph ( kamiski et al . , 2014 ) .
espero is designed to be fiber - fed from the ritchey - chretien ( rc ) focus of the 2-m telescope .
figure [ fig : dome ] shows the overall schema of the spectrograph design .
a fiber injection and guiding unit ( figu ) is mounted at the rc - focus .
the calibration unit , containing the calibration lamps ( led , tungsten and thar ) is placed in the control room at the level of the telescope .
the spectrograph itself is in a separate , thermally isolated room , one level below the telescope , next to the coude spectrograph room .
[ fig : dome ] the light is transmitted from the rc focus to the spectrograph via an optical silica / silica multi - mode fiber with 50 @xmath2 m core , 30 m length , and numerical aperture ( n.a . ) = 0.22 .
the entrance of the fiber is mounted on the fiber injection and guiding unit ( figu ) , produced by shelyak instruments .
detailed description of figu can be found on the website of shelyak instruments .
the main function of figu is to inject the focused light of the observed object into the object fiber . but this versatile unit is used , also , to transfer light from the calibration unit ( for flat fielding ( tungsten lamp and led ) and for wavelength calibration ( thar lamp ) ) to the object fiber , and as well to transfer some amount of the object s light to a guiding ccd camera
. the light beam from a star is focused by the telescope on a hole in the center of an inclined mirror .
the central part of the image is transmitted through the hole , and the wings are reflected to the guiding camera .
the calibration unit , comprising of the thar lamp for wavelength reference and the tungsten and led for flat fieldig , is placed in a separate room at the level of the telescope .
the calibration unit was also purchased from shelyak . in order to achieve best match between the telescope focal ratio and the image scale ( f/8 and 77.5 @xmath2m/ , respectively ) with that of the figu
, we purchased the f/9 version , with a 75 @xmath2 m hole in the mirror .
an optical system below the mirror , having transfer ratio 1.5:1 , refocusses the 75 @xmath2 m hole onto the tip of the 50 @xmath2 m object fiber , and the focal ratio is changed from f/8 to f/5.3 , which corresponds to a n.a .
= 0.09 , far below that of the fiber .
the light beam from the calibration unit is transferred to the figu via a 200 @xmath2 m fiber .
an automatically controlled moving flat mirror reflects the beam into the hole of the mirror , and from there on the light follows the same optical path to the spectrograph as the beam from the object .
the autoguiding is performed with an atik titan camera , which comprises of a ccd chip with 659 x 494 quadratic pixels of size 7.4 @xmath2 m .
detailed description of the camera can be found on the vendor s webpage . before reaching the ccd , the reflected image of the object passes through a transfer optics with 1.5x magnification .
thus , on the guiding ccd , 1@xmath3 corresponds to 120 @xmath2 m , the scale is 0.06@xmath4 m , and the fov = 40@xmath3 x 30@xmath3 .
a dedicated mechanical mounting was manufactured to hold all the components in the rc - focus , with special attention to the stable fixation of both fibers , eliminating the mechanical load on their connectors at different positions of the telescope .
figure [ fig : ontable ] shows the instrument in the spectrograph room .
the fiber feeding of the spectrograph is marked with number 1 .
the exit of the fiber is mounted in the housing of an optical adapter ( label 2 in fig .
[ fig : ontable ] , which serves for adjustment of the output beam to the f/15 beam of the collimator .
this adaptation optics is just behind the slit ( number 3 in fig .
[ fig : ontable ] ) .
the output end of the object fiber with its optical adapter and the slit holder are firmly fixed on an originally designed and manufactured in nao table with five orders of freedom .
this mechanical device is aimed to ensure proper adjustment of incoming light beam onto the collimator focal plane .
the components 4,5,6,7 , and 8 are the off - axis collimator ( oap ) , the echelle grating , the cross - dispersing prism , the camera lens , and the detector , respectively .
the oap is an off - the - shelf part purchased from optical surfaces ltd .
it has a focal length of 1508.2 mm , an off - axis distance 103.9 mm , and a free aperture of 140 mm .
our optimized optical scheme is based on a parallel beam of 100 mm , thus we are working far from the edges of the mirror with an effective focal ratio of f/15 . the r2 echelle grating has 37.5 lines / mm , its dimensions are 240x120x30 mm , and the blaze angle is 63.5@xmath5 .
it works in a nearly littrow regime , with an angle of 12@xmath5 between the plane of dispersion and the directon to the crossdispersing prism .
the crossdisperser is a prism made of glass f108 , having characteristics similar to the f7 glass of schott .
the apex angle of the prism is 53.7@xmath5 and the dimensions of its entrance surface are 187x190 mm .
the imaging camera is a commercial canon ef 400 mm f/2.8l is usm lens .
this objective lens has a diagonal fov of 6@xmath510@xmath6 , which is well adapted to the size of the detector ( see next subsection ) .
the detector is an ikon - l ccd camera , manufactured by andor .
it uses a thermoelectric cooled , back - illuminated , scientific - grade e2v ccd42 - 40 chip , comprising 2048x2048 quadratic pixels of size 13.5 @xmath2 m .
the camera is characterized by low readout noise , high sensitivity and high dynamic range .
when reading with 50 khz the gain is 1 e@xmath7/adu , and the ron is 3.4 electrons .
the digitization is performed by a 16-bit analogue to digital converter , and the saturation level is over 88000 electrons .
all components of the spectrograph are mounted on a honeycomb optical table with dimensions 2400 x 1200 mm , delivered by standa .
the table itself stays on 4 pneumatic vibration isolators , which keep the level of the table and damp resonances caused by ambient vibrations .
the vibration isolators are permanently connected to a cylinder of pressurized nitrogen . for better thermal isolation and stability and in order to keep the spectrograph free of dust and other possible contaminants ,
it is enclosed in a non - hermetic box .
all components of the spectrograph , except the ccd , are in the enclosure .
this way any effect of the detector ventilation on the other parts of the spectrograph is avoided . to let the ccd outside ,
the box was manufactured with one hole between the end of the camera lens and the opening of the ccd .
the ccd detector is operated by an industrial pc running solis , proprietary software supplied by andor together with the camera .
because of the cable length limits of the usb connection , the pc is located relatively close to the camera , in the spectrograph room .
three additional single board computers ( sbcs ) operate ( 1 ) the setting of the slit width , ( 2 ) the monitoring of the temperature inside of the enclosure and in the spectrograph room environment , and ( 3 ) the autoguiding of the telescope during exposures .
the slit width is controlled by an arduino uno board located in the spectrograph room .
this board is connected via rs-232 interface to the pc that operates the ccd camera , this way the slit width can be controlled remotely over the local area network ( lan ) with a simple software tool developed for the purpose .
the temperature and humidity monitoring is done by a raspberry pi b sbc , which reads , records the sensor readings and provides web interface to the stored data .
there are five sensors connected : three located in the spectrograph enclosure ( on the ccd camera , on the collimator and on the grating ) and two in the spectrograph room ( on the ceiling and at the optical table ) .
the monitoring of the temperature over several months in 2016 gives a mean temperature of [email protected]@xmath5c .
the relative humidity shows larger fluctuations , between 30% and 50% .
these values resulted from a very preliminary analysis of the monitoring data , and they give enough information about the future needs of better stabilisation of the spectrograph environment , especially of the humidity . for the autoguiding another sbc
is used , raspberry pi b2 , which runs lin_guider .
lin_guider is a free software originally developed for amateur telescope guiding , but extended with new features which made it a professional - grade software .
it is being actively developed and extended with new features to make it more suitable for guiding on the residual stellar image as the majority of the light goes into the object optical fibre and the guiding is performed only based on the tiny percentage of the side lobe light .
lin_guider currently uses centroid algorithm , but tests are being performed with two new algorithms which will be part of the next major release .
both of them can be used for full frame guiding and all three have a subpixel precision .
however each of them offers different benefits .
one of them : donuts ( mccormac et al.,2013 ) is suitabe for guiding on defocussed stellar image , useful for high precision photometry . and the second is specially designed for guiding using the side lobes of the stellar image , which is basically a stellar image with a moving hole in it , hence its name holes
. the holes algorithm will be descibed in detail in a separate paper . the guiding camera , atik titan , is equipped with standard st4 guiding port .
for that purpose a special adapter is created to make it possible to control the 2_meter telescope via the st4 port .
this approach has another benefit - any of the wide range of cameras , supported by lin_guider , that is equipped with guiding port , can be used instead . the detector pc and the three sbcs are controlled remotely over the lan .
thus , entering the spectrograph room is necessary only on exceptional occasions .
for calculation of the orders and their corresponding wavelengths we used the grating equation , as given by schroeder and hilliard ( 1980 ) : @xmath9 where @xmath10 is the order of the wavelength @xmath1 which is diffracted from an echelle grating of grove spacing @xmath11 , and @xmath12 , @xmath13 , and @xmath14 are the angles of incidence , diffraction and deviation from the plane defined by the incident and diffracted rays , respectively .
cccccccc order & @xmath15&@xmath16&@xmath17&@xmath17-@xmath15&fsr&@xmath17-@xmath15-fsr&dispersion + & & & & & & & / px + + 53 & 8849.97 & 8927.87 & 9000.66 & 150.69 & 168.45 & -17.76 & 0.0736 + 54 & 8686.08 & 8762.54 & 8833.98 & 147.90 & 162.27 & -14.37 & 0.0722 + 55 & 8528.15 & 8603.22 & 8673.36 & 145.21 & 156.42 & -11.21 & 0.0709 + 56 & 8375.86 & 8449.59 & 8518.48 & 142.62 & 150.89 & -8.27 & 0.0696 + 57 & 8228.92 & 8301.36 & 8369.03 & 140.11 & 145.64 & -5.52 & 0.0684 + 58 & 8087.04 & 8158.23 & 8224.74 & 137.70 & 140.66 & -2.96 & 0.0672 + 59 & 7949.97 & 8019.95 & 8085.34 & 135.36 & 135.93 & -0.57 & 0.0661 + 60 & 7817.47 & 7886.29 & 7950.58 & 133.11 & 131.44 & 1.67 & 0.0650 + 61 & 7689.32 & 7757.00 & 7820.24 & 130.93 & 127.16 & 3.76 & 0.0639 + 62 & 7565.30 & 7631.89 & 7694.11 & 128.81 & 123.10 & 5.72 & 0.0629 + 63 & 7445.21 & 7510.75 & 7571.98 & 126.77 & 119.22 & 7.55 & 0.0619 + 64 & 7328.88 & 7393.40 & 7453.67 & 124.79 & 115.52 & 9.27 & 0.0609 + 65 & 7216.13 & 7279.65 & 7339.00 & 122.87 & 111.99 & 10.88 & 0.0600 + 66 & 7106.79 & 7169.35 & 7227.80 & 121.01 & 108.63 & 12.38 & 0.0591 + 67 & 7000.72 & 7062.35 & 7119.92 & 119.20 & 105.41 & 13.79 & 0.0582 + 68 & 6897.77 & 6958.49 & 7015.22 & 117.45 & 102.33 & 15.12 & 0.0573 + 69 & 6797.80 & 6857.64 & 6913.55 & 115.75 & 99.39 & 16.36 & 0.0565 + 70 & 6700.69 & 6759.68 & 6814.78 & 114.09 & 96.57 & 17.53 & 0.0557 + 71 & 6606.32 & 6664.47 & 6718.80 & 112.49 & 93.87 & 18.62 & 0.0549 + 72 & 6514.56 & 6571.91 & 6625.49 & 110.92 & 91.28 & 19.65 & 0.0542 + 73 & 6425.32 & 6481.88 & 6534.73 & 109.40 & 88.79 & 20.61 & 0.0534 + 74 & 6338.49 & 6394.29 & 6446.42 & 107.93 & 86.41 & 21.52 & 0.0527 + 75 & 6253.98 & 6309.03 & 6360.47 & 106.49 & 84.12 & 22.37 & 0.0520 + 76 & 6171.69 & 6226.02 & 6276.78 & 105.09 & 81.92 & 23.16 & 0.0513 + 77 & 6091.54 & 6145.16 & 6195.26 & 103.72 & 79.81 & 23.91 & 0.0506 + 78 & 6013.44 & 6066.38 & 6115.83 & 102.39 & 77.77 & 24.62 & 0.0500 + 79 & 5937.32 & 5989.59 & 6038.42 & 101.10 & 75.82 & 25.28 & 0.0494 + 80 & 5863.10 & 5914.72 & 5962.94 & 99.83 & 73.93 & 25.90 & 0.0487 + 81 & 5790.72 & 5841.69 & 5889.32 & 98.60 & 72.12 & 26.48 & 0.0481 + 82 & 5720.10 & 5770.46 & 5817.50 & 97.40 & 70.37 & 27.03 & 0.0476 + 83 & 5651.19 & 5700.93 & 5747.41 & 96.22 & 68.69 & 27.54 & 0.0470 + 84 & 5583.91 & 5633.06 & 5678.99 & 95.08 & 67.06 & 28.02 & 0.0464 + 85 & 5518.22 & 5566.79 & 5612.18 & 93.96 & 65.49 & 28.47 & 0.0459 + 86 & 5454.05 & 5502.06 & 5546.92 & 92.87 & 63.98 & 28.89 & 0.0453 + 87 & 5391.36 & 5438.82 & 5483.16 & 91.80 & 62.52 & 29.28 & 0.0448 + 88 & 5330.10 & 5377.01 & 5420.85 & 90.76 & 61.10 & 29.65 & 0.0443 + 89 & 5270.21 & 5316.60 & 5359.94 & 89.74 & 59.74 & 30.00 & 0.0438 + 90 & 5211.65 & 5257.53 & 5300.39 & 88.74 & 58.42 & 30.32 & 0.0433 + 91 & 5154.38 & 5199.75 & 5242.14 & 87.76 & 57.14 & 30.62 & 0.0429 + 92 & 5098.35 & 5143.23 & 5185.16 & 86.81 & 55.90 & 30.90 & 0.0424 + 93 & 5043.53 & 5087.93 & 5129.41 & 85.88 & 54.71 & 31.17 & 0.0419 + 94 & 4989.88 & 5033.80 & 5074.84 & 84.96 & 53.55 & 31.41 & 0.0415 + 95 & 4937.35 & 4980.81 & 5021.42 & 84.07 & 52.43 & 31.64 & 0.0410 + 96 & 4885.92 & 4928.93 & 4969.11 & 83.19 & 51.34 & 31.85 & 0.0406 + 97 & 4835.55 & 4878.12 & 4917.89 & 82.34 & 50.29 & 32.05 & 0.0402 + 98 & 4786.21 & 4828.34 & 4867.70 & 81.50 & 49.27 & 32.23 & 0.0398 + 99 & 4737.86 & 4779.57 & 4818.53 & 80.67 & 48.28 & 32.39 & 0.0394 + 100 & 4690.48 & 4731.77 & 4770.35 & 79.87 & 47.32 & 32.55 & 0.0390 + 101 & 4644.04 & 4684.92 & 4723.12 & 79.07 & 46.39 & 32.69 & 0.0386 + 102 & 4598.51 & 4638.99 & 4676.81 & 78.30 & 45.48 & 32.82 & 0.0382 + 103 & 4553.87 & 4593.95 & 4631.41 & 77.54 & 44.60 & 32.94 & 0.0379 + 104 & 4510.08 & 4549.78 & 4586.87 & 76.79 & 43.75 & 33.05 & 0.0375 + 105 & 4467.13 & 4506.45 & 4543.19 & 76.06 & 42.92 & 33.14 & 0.0371 + 106 & 4424.98 & 4463.94 & 4500.33 & 75.34 & 42.11 & 33.23 & 0.0368 + 107 & 4383.63 & 4422.22 & 4458.27 & 74.64 & 41.33 & 33.31 & 0.0364 + 108 & 4343.04 & 4381.27 & 4416.99 & 73.95 & 40.57 & 33.38 & 0.0361 + 109 & 4303.20 & 4341.08 & 4376.47 & 73.27 & 39.83 & 33.44 & 0.0358 + 110 & 4264.08 & 4301.61 & 4336.68 & 72.60 & 39.11 & 33.50 & 0.0355 + 111 & 4225.66 & 4262.86 & 4297.61 & 71.95 & 38.40 & 33.55 & 0.0351 + 112 & 4187.93 & 4224.80 & 4259.24 & 71.31 & 37.72 & 33.59 & 0.0348 + 113 & 4150.87 & 4187.41 & 4221.55 & 70.68 & 37.06 & 33.62 & 0.0345 + 114 & 4114.46 & 4150.68 & 4184.52 & 70.06 & 36.41 & 33.65 & 0.0342 + 115 & 4078.68 & 4114.58 & 4148.13 & 69.45 & 35.78 & 33.67 & 0.0339 + 116 & 4043.52 & 4079.11 & 4112.37 & 68.85 & 35.16 & 33.68 & 0.0336 + 117 & 4008.96 & 4044.25 & 4077.22 & 68.26 & 34.57 & 33.69 & 0.0333 + 118 & 3974.99 & 4009.98 & 4042.67 & 67.68 & 33.98 & 33.70 & 0.0330 + 119 & 3941.58 & 3976.28 & 4008.70 & 67.11 & 33.41 & 33.70 & 0.0328 + 120 & 3908.74 & 3943.14 & 3975.29 & 66.55 & 32.86 & 33.69 & 0.0325 + [ tab : orders ] table [ tab : orders ] lists the distribution of wavelengths among the different orders .
a total of 71 orders cover the spectral range from 3813 to 9000 .
the titles of the columns in that table explain their content .
in the sixth column , the negative values in the first seven rows are the gaps in the corresponding orders .
the separation of the orders over the ccd chip is achieved with putting the cross - disperser in the beam of diffracted light .
figure [ fig : orders ] shows the distribution of the orders over the detector . for clarity , every fifth order is given , and marked with the order number and its wavelength limits .
evidently the ccd is effectively used , the orders are spread over the full frame of the chip .
the extraction of all orders in the obtained spectra requires a well exposed set of flat fields , spectra of hot stars or solar spectra , giving sufficient signal over the whole range of wavelengths .
the extracted orders are listed in tab .
[ tab : apertures ] .
the central wavelength and dispersion were derived from wavelength calibration with the thar lines , identified in every order .
the separation between the extracted extremely blue orders amounts to 47 pixels and between the extremely red orders - 17 pixels .
ccccc aperture & order & row center & @xmath16 & order dispersion + number & number & pixel & & / px + + 2 & 53 & 66.4 & 8939.8 & 0.0771 + 3 & 54 & 83.1 & 8774.3 & 0.0756 + 4 & 55 & 99.8 & 8614.8 & 0.0743 + 5 & 56 & 116.7 & 8460.9 & 0.0729 + 6 & 57 & 133.8 & 8312.5 & 0.0717 + 7 & 58 & 151.0 & 8169.2 & 0.0704 + 8 & 59 & 168.5 & 8030.7 & 0.0693 + 9 & 60 & 186.2 & 7896.9 & 0.0681 + 10 & 61 & 204.1 & 7767.4 & 0.0670 + 11 & 62 & 222.2 & 7642.2 & 0.0659 + 12 & 63 & 240.7 & 7520.9 & 0.0649 + 13 & 64 & 259.3 & 7403.4 & 0.0639 + 14 & 65 & 278.2 & 7289.5 & 0.0629 + 15 & 66 & 297.3 & 7179.0 & 0.0619 + 16 & 67 & 316.8 & 7071.9 & 0.0610 + 17 & 68 & 336.5 & 6967.9 & 0.0601 + 18 & 69 & 356.5 & 6866.9 & 0.0592 + 19 & 70 & 376.7 & 6768.8 & 0.0584 + 20 & 71 & 397.2 & 6673.5 & 0.0576 + 21 & 72 & 418.1 & 6580.8 & 0.0568 + 22 & 73 & 439.3 & 6490.7 & 0.0560 + 23 & 74 & 460.8 & 6403.0 & 0.0553 + 24 & 75 & 482.6 & 6317.6 & 0.0545 + 25 & 76 & 504.8 & 6234.5 & 0.0538 + 26 & 77 & 527.3 & 6153.5 & 0.0531 + 27 & 78 & 550.1 & 6074.6 & 0.0524 + 28 & 79 & 573.3 & 5997.8 & 0.0518 + 29 & 80 & 596.9 & 5922.8 & 0.0511 + 30 & 81 & 620.8 & 5849.7 & 0.0505 + 31 & 82 & 645.1 & 5778.3 & 0.0499 + 32 & 83 & 669.8 & 5708.7 & 0.0493 + 33 & 84 & 694.9 & 5640.8 & 0.0487 + 34 & 85 & 720.2 & 5574.4 & 0.0481 + 35 & 86 & 746.1 & 5509.6 & 0.0476 + 36 & 87 & 772.3 & 5446.3 & 0.0470 + 37 & 88 & 799.0 & 5384.4 & 0.0465 + 38 & 89 & 826.1 & 5323.9 & 0.0460 + 39 & 90 & 853.7 & 5264.8 & 0.0455 + 40 & 91 & 881.7 & 5206.9 & 0.0450 + 41 & 92 & 910.1 & 5150.3 & 0.0445 + 42 & 93 & 939.0 & 5095.0 & 0.0440 + 43 & 94 & 968.2 & 5040.8 & 0.0435 + 44 & 95 & 998.1 & 4987.7 & 0.0431 + 45 & 96 & 1028.4 & 4935.8 & 0.0426 + 46 & 97 & 1059.2 & 4884.9 & 0.0422 + 47 & 98 & 1090.6 & 4835.0 & 0.0417 + 48 & 99 & 1122.3 & 4786.2 & 0.0413 + 49 & 100 & 1154.8 & 4738.4 & 0.0409 + 50 & 101 & 1187.7 & 4691.5 & 0.0405 + 51 & 102 & 1221.1 & 4645.5 & 0.0401 + 52 & 103 & 1255.1 & 4600.4 & 0.0397 + 53 & 104 & 1289.7 & 4556.1 & 0.0393 + 54 & 105 & 1324.9 & 4512.8 & 0.0389 + 55 & 106 & 1360.7 & 4470.2 & 0.0386 + 56 & 107 & 1397.0 & 4428.4 & 0.0382 + 57 & 108 & 1434.1 & 4387.4 & 0.0379 + 58 & 109 & 1471.8 & 4347.2 & 0.0375 + 59 & 110 & 1510.1 & 4307.7 & 0.0372 + 60 & 111 & 1549.1 & 4268.9 & 0.0368 + 61 & 112 & 1588.9 & 4230.8 & 0.0365 + 62 & 113 & 1629.2 & 4193.3 & 0.0362 + 63 & 114 & 1670.5 & 4156.6 & 0.0358 + 64 & 115 & 1712.4 & 4120.4 & 0.0355 + 65 & 116 & 1755.1 & 4084.9 & 0.0352 + 66 & 117 & 1798.8 & 4050.0 & 0.0349 + 67 & 118 & 1843.9 & 4015.7 & 0.0346 + 68 & 119 & 1886.6 & 3982.0 & 0.0343 + 69 & 120 & 1934.0 & 3948.8 & 0.0340 + [ tab : apertures ] comparison of the central wavelengths of the extracted orders with the calculated ones shows small differences which are gradually decreasing from about 12 in order 53 ( 8940 ) to below 6 in order 120 ( 3950 ) .
this dependence is presented in fig .
[ fig : o_c ] .
the relative values of these differences are almost constant over the entire spectral range .
the role of high quality flat fields is very important in the process of extraction of orders .
the flat field must not only correct for the pixel - to - pixel sensitivity , but it has also to compensate the blaze of the echelle grating and remove the fringing in the red spectral region . in order to fulfil all these functions at the same time it is necessary to have sufficient signal over the entire spectral range .
it is hard to meet this requirement , especially for the blue orders where the signal of the flat field is very low .
this situation is well illustrated on fig .
[ fig : tung - led ] , where a profile of a flat field , taken in the cross - dispersion direction , is shown .
the blue end is at the right , and it is seen that the intensity of the led drops very fast in that direction .
the decreasing quantum efficiency of the ccd in the blue spectral region additionally contributes for lower signal in that region . in order to obtain sufficient signal in the last blue orders it is necessary to make flat fields with longer exposures , and subsequently to combine the flats with different exposures into an usable superflat .
another feature of the led and tungsten lamp is that their spectra , considered in a given single order , are inconsistent . probably , the reason is the different color temperature of the tungsten and the led .
up to date it is not clear which of the lamps better removes the echelle blaze function , therefore we recommend to expose the tungsten and led lamps separately , and find empirically which lamp gives better results for every particular order . as mentioned above the extraction of orders
depends on high - quality flat fields .
this dependence is especially critical in the red spectral region , where fringe patterns start to appear close to h@xmath12 and are getting stronger with increasing wavelength . for wavelengths greater than 7000
the fringing is so strong that it can compromise the scientific information contained in the spectra , if they are not processed correctly with suitable flat fields .
an example of removal of fringes in the range 8380 - 8530 is presented in fig .
[ fig : fringes ] .
the final result , presented in the lower panel of that figure , shows that after the division by the flat field the fringe patterns are excellently removed , and the blaze function of the grating and the influence of the pixel - to - pixel sensitivity are eliminated simultaneously .
one example of the derived resolving power of the spectrograph ( r = @xmath1/@xmath18 ) is presented in fig .
[ fig : noslit_resolution ] .
the full width at half maximum ( fwhm = @xmath18 ) of more than thousand lines distributed over all orders were measured by fitting the lines with gaussian function . some of the lines , especially in the blue region , appear in adjacent orders .
these lines are measured twice , but the contribution of every measurement has the same weight , as in these cases the measurements are independent . .
[ fig : noslit_resolution ] there is a possibility to obtain spectra with higher resolutions by decreasing the width of the slit , placed in the oap collimator focal plane .
of course , going to smaller widths , after a given value , a decrease of the signal will be experienced . in order to estimate the expected losses ,
the fwhm of calibration lines , selected within five randomly distributed orders , were measured .
the result of these measurements are shown in fig .
[ fig : slit - fwhm ] . here
, the upper abscissa shows the width at which the slit is set , and the lower one is its projection on the chip .
the projection is calculated by the relation describing the transfer through the spectrograph optics ( @xmath19 ) , where _ w _ is the slit width in microns , @xmath20 is the ccd pixel size(@xmath21 ) , and _
k _ is the spectrograph transmission factor ( @xmath22 , where _ gm _ is the grating magnification factor , _ fcol _ and _ fcam _ are the oap and canon focal lengths , respectively ) .
the most remarkable feature in this figure is the constant fwhm for slit widths greater than 200 @xmath2 m , which is an indication that over this limit the spectra are obtained with the maximum possible illumination .
the relatively uniform distribution of r over the whole spectral range was achieved after careful focusing , including changes in the distance and the tilt between the camera lens and the detector . to check the uniformity of the optical alignment we measured the full widths of many thar spectrum lines situated in different spectral regions .
these regions are shown on fig .
[ fig : ccd_rgb ] and are provisionally denoted as blue , green and red .
figure [ fig : fwhm_cols ] shows the fwhms distribution along columns ( dispersion ) .
no significant trends either along columns or rows can be seen , what means that the focal plane of the lens is well - aligned with the plane of the ccd chip . to evaluate the signal loss at smaller slit width twenty three thorium - argon lamp echellograms
were taken , gradually changing the width of the slit from @xmath23 to @xmath24 , and keeping the exposure constant .
for every echellogram ( corresponding to a certain slit width as well ) a single ( unblended ) unsaturated thorium - argon emission line was gaussian approximated within five randomly chosen orders .
this way twenty three flux values corresponding to different slit widths were derived .
the change of these flux values along echelograms was accepted as a measure of flux loss related to the decrease of the slit width . for every
emission line gaussian flux parameter values were normalized to maximum flux value in that sample ( @xmath25 ) .
the normalized flux values in percentage corresponding to the five lines in every echellogram were averaged to give a mean flux value at every slit width .
the mean flux standard deviation was calculated also and its value accepted as an error of the flux loss measurement .
to every flux loss value we attached a magnitude , following @xmath26 .
these data are listed in tab . [
tab : losses ] .
column ( 1 ) presents the slit width , columns ( 2 ) and ( 3 ) contain the flux loss and its uncertainty in percentage , columns(4 ) and ( 5 ) present the same in terms of stellar magnitudes as defined above , columns ( 6 ) and ( 7 ) contain the fwhm of the gaussian fit and its uncertainties .
cccccccc slit & & flux & stddev & magnitude & stddev & fwhm & fwhm err + um & & % & % & mag & mag & px & px + + 50 & & 76.8 & 1.3 & 1.59 & 0.05 & 1.69 & 0.08 + 60 & & 71.2 & 1.3 & 1.36 & 0.04 & 1.82 & 0.08 + 70 & & 66.6 & 1.5 & 1.19 & 0.05 & 1.94 & 0.05 + 80 & & 60.6 & 1.5 & 1.01 & 0.04 & 2.08 & 0.04 + 90 & & 56.0 & 1.7 & 0.89 & 0.04 & 2.13 & 0.07 + 100 & & 50.6 & 2.1 & 0.76 & 0.04 & 2.28 & 0.05 + 110 & & 44.6 & 2.6 & 0.65 & 0.05 & 2.39 & 0.07 + 120 & & 40.2 & 0.8 & 0.56 & 0.01 & 2.53 & 0.07 + 130 & & 35.2 & 1.3 & 0.47 & 0.02 & 2.70 & 0.04 + 140 & & 31.6 & 1.7 & 0.41 & 0.02 & 2.83 & 0.05 + 150 & & 25.4 & 1.5 & 0.32 & 0.02 & 2.95 & 0.06 + 160 & & 20.6 & 1.5 & 0.25 & 0.02 & 3.08 & 0.05 + 170 & & 18.2 & 0.8 & 0.22 & 0.01 & 3.21 & 0.05 + 180 & & 15.8 & 0.8 & 0.18 & 0.01 & 3.29 & 0.04 + 190 & & 11.2 & 0.4 & 0.13 & 0.01 & 3.43 & 0.04 + 200 & & 9.2 & 2.2 & 0.11 & 0.03 & 3.54 & 0.05 + 210 & & 5.6 & 0.9 & 0.06 & 0.01 & 3.63 & 0.06 + 220 & & 4.2 & 1.1 & 0.05 & 0.01 & 3.67 & 0.06 + 230 & & 1.6 & 1.1 & 0.02 & 0.01 & 3.77 & 0.04 + 240 & & 2.0 & 1.0 & 0.02 & 0.01 & 3.79 & 0.07 + 250 & & 4.0 & 2.3 & 0.05 & 0.03 & 3.81 & 0.07 + 300 & & 0.0 & 0.0 & 0.00 & 0.00 & 3.87 & 0.08 + 350 & & 1.8 & 1.3 & 0.02 & 0.01 & 3.84 & 0.10 + [ tab : losses ] the upper panel in fig .
[ fig : flux_mag_loss ] shows the flux loss ( in percentages ) vs slit width , and the lower panel shows the flux loss in terms of magnitude .
another illustration of a possible change of the fwhm of the spectral lines with variations of the slit width is presented on fig .
[ fig : slit_project ] . here , we use only the three lines marked on fig . [
fig : ccd_rgb ] . a subimage ( @xmath87 x @xmath85 px along the horizontal and vertical direction , respectively )
is extracted around the center of every line , and transformed to 1d , by averaging along the vertical direction .
the 1d - line was fitted with a gauss function , and the fwhm was derived from the parameters of the fit .
the comparison between the measured fwhm and the calculated projection of the slit width shows that no increase of the fwhm is experienced for slit widths greater than about 220 @xmath2 m .
this can be considered as a measure of the diameter of the beam image in the focal plane of the collimator ( the plane of the slit ) . in the region of very small slit widths
the fwhm , as expected , approximates gradually the limit of 2 px ( nyquist theorem ) .
the measurements of the three lines marked on fig .
[ fig : ccd_rgb ] were further used to assess together both , the increase in resolving power and the decrease in flux when narrowing the slit width .
these dependencies are presented on fig .
[ fig : resolve_power ] . as a measure of the decreased flux we used the maxima of the gaussian fit , described in the previous paragraph .
the scale for the resolving power is on the right ordinate . for slit widths greater than about 200 @xmath2 m no substantial changes of the flux and resolution
are observed , because this is equivalent to taking spectra without slit . for slit widths lower than about 110 @xmath2 m ( corresponding to @xmath272 px projection on the ccd ) the flux losses dominate over the increase in resolution ,
thus it is not recommended to close the slit below this limit .
a good measure of the efficiency of the spectrograph is the signal - to - noise ratio ( snr ) obtained at different illumination levels .
figure [ fig : snr - mag ] shows the derived snr per pixel as a function of the stellar magnitude .
this relation is a compilation of three clear observing nights with more than 8 observed objects per night and seeing 1 - 2 .
the seeing is evaluated from stellar image projection on the fiber entrance placed in the telescope focal pane .
the snr measurements of the nights with seeing 1 - 1.5were transformed to the results from the night with 2seeing using least square method .
stars of different magnitudes were observed with different exposures , but for the final set of data they were all transformed to exposure of 1800 s. the straight line presents linear regression taken from the snr measurements of the three nights .
spectra taken at airmass @xmath28 2 were not used .
stellar magnitudes of the objects were reduced to unity airmass using extinction coefficient of 0.2 , a value commonly accepted at nao rozhen .
the dashed lines in fig . [ fig : snr - mag ] , above and below the empirically derived dependence , show the expected signal - to - noise ratio in case of seeing 1and 3 , respectively .
these estimations were made with a 2d gaussian function , calculated for different seeing sizes , and measuring the portion of the flux proportional to the cross - section of the entrance tip of the object fiber . from the data summarized in fig .
[ fig : snr - mag ] , for snr=100 and exposure 1800 s , the limiting magnitudes are 8.8 , 7.5 , and 6.8 , for seeing 1 , 2 , and 3 , respectively .
in order to determine the zero point of the radial velocities ( rv ) derived from espero spectra observations of appropriate standard stars were obtained every observing night .
stars were selected mainly from the gaia catalog of rv standard stars ( soubiran et al . , 2013 ) .
sixty nine spectra of twenty three standard stars were obtained in the period june 2015 - september 2016 .
rvs were obtained by crosscorrelating the spectrum of the studied object with a template spectrum ( template ) .
templates were generated using spectrum , a stellar spectral synthesis program written by gray and corbally ( 1994 ) , and pollux , a stellar spectra database offering access to theoretical data ( palacios et al . , 2010 ) .
every rv is derived as a mean value of measurements obtained from many orders .
orders with a low signal - to - noise ratio as well as strong telluric absorption features were omitted .
figure [ fig : rv - monitor ] presents the difference @xmath29 radial velocity value .
the annual monitoring of this difference shows slight systematic deviation of measured velocities with a mean value of -0.18 km s@xmath30 with a standard deviation of the mean 0.07 km s@xmath30 .
more thorough analysis of the radial velocity accuracy derived from espero and its dependence on spectrograph working conditions will be discussed in a forthcoming paper .
fig15.eps the coverage of the full spectral range of the spectrograph is best illustrated after merging all orders into a single 1-d spectrum . here again , as described in section [ subsec : identi_ord ] , the good flat field for every order is very important , to remove the blaze function of the echelle grating .
some residuals after dividing the orders by their corresponding flat fields can be removed by using spectra of standard stars .
spectrophotometric standard - star observations are also needed to derive the response function of the spectrograph .
one example of a 2-d echellogram transformed into an 1-d spectrum , and calibrated in relative fluxes is given in fig .
[ fig : hr718_flux ] . given the described characteristics of the spectrograph
it is interesting to compare spectra of espero with spectra of the same objects , obtained with echelle spectrograph used on a 2-meter class telescope . in fig .
[ fig : epsaur_comp ] part of the spectrum of @xmath31 aur from the elodie database ( moultaka et al .
2004 ) is compared with one obtained with espero .
the nominal resolution of elodie is about 42000 , and the spectrum obtained with espero is exposed without slit , i.e. the r is about 30000 .
nevertheless no substantial differences are seen between both spectra .
the first published result , based on data obtained with espero is presented in bonev et al .
( 2014 ) . in this telegram
the authors report on the most prominent features measured in a spectrum of sn2014j on january 23.75 ut , 2014 .
stoyanov and zamanov ( 2016 ) present optical spectroscopy of the high - mass x - ray binary a0535 + 26 ( hde 245770 , v725 tau ) after the periastron passage .
zamanov et al .
( 2016 ) describe spectral observations of four be / x - ray and @xmath14-ray binaries .
tomov et al .
( 2016 ) studied the symbiotic star ag peg during its 2015 outburst .
espero is now operational . here
we presented its design , performance and working status .
several improvements can enhance its efficiency in future .
so , e.g. a focal reducer could be used in the rc - focus for better matching between the typical seeing at rozhen and the parameters of the fiber injection unit .
alternative methods for obtaining flat fields better matching the science frames ( one option is to design a system for obtaining dome flats ) . with a specially designed filter the illumination of the tungsten lamp ( and led ) could be modified to have a rather flat spectral distribution .
new algorithms for more efficient autoguiding are under development .
replacement of the cross - disperser with a new one aiming to improve light transmission in the blue spectral range . a new software dedicated to control spectrograph functions and to complete fits headers with a relevant astronomical information during the observing night is still needed . _
acknowledgements : _ we are thankful to remi cabanac and his team at pic du midi observatory and tarbes for the extensive and valuable discussions concerning the design of the narval spectrograph , including its budget and prices of components . in toulouse , laurent pares provided us very detailed information about the optical design of echelle spectrographs in general , and about espadons in particular .
the plant for optics in panagyurishte played a very important role in the process of assembling and alignment of the spectrograph .
this research made use of the pollux database operated at lupm ( universit montpellier - cnrs , france with the support of the pnps and insu . ) .
the preliminary processing and the data analysis was performed mainly with iraf routines .
high resolution stellar spectra were downloaded from the elodie archive in order to be compared with spectra obtained with espero .
the development of the design of the spectrograph and its manufacturing were financially supported by contracts do 02 - 85 and do 1 - 34 with the ministry of science and education .
the authors would like to thank the the bulgarian academy of sciences for the support for contracts dfnp-103/11.05.2016 , dfnp-106/11.05.2016 , dfnp-107/11.05.2016 , dfnp-122/11.05.2016 `` support for young scientists '' .
bonev , t. , tomov , t. , swierczynski , e. , iliev , i. , dimitrov , d. , markov , h. , stoyanov , k. , belcheva , m. , nikolov , g. , nikolov , p. , chanliev , d. , churalski , m. , nikolov , y. , kurtenkov , a. , stateva , i. , petrov , n. , dimitrov , w. , musaev , f. , tsvetanov , z. , miloushev , i. , tenev , t. , 2014 , _ the astronomer s telegram , v. 5829 _ | in this paper we describe the echelle spectrograph of the 2 meter telescope of the rozhen national astronomical observatory .
the spectrograph is a cross - dispersed , bench - mounted , fiber - fed instrument giving a resolution from @xmath030000 to @xmath045000 .
the spectral range obtained in one single image is from 3900 to 9000 .
we describe the parameters of the fiber injection and the guiding unit , of the spectrograph itself , and of the detector .
the identified orders and the resulting resolving power are presented . the opportunity to increase of the resolution by using a narrower slit
is discussed and the corresponding loss of flux is calculated .
the expected signal - to - noise ratio for a set of stars of different magnitudes was derived .
some of the first results obtained with espero are shortly described . |
recently , majorana s stellar representation ( msr ) has gained revived interest as it offers an intuitive geometrical approach to understanding quantum systems with multiple components by mapping states in the higher dimensional space as points on a bloch sphere @xcite . in general , a quantum state of spin-1/2 systems or equivalently two - level systems can be represented as a point on a unit sphere @xcite , and the evolution of the quantum state corresponds to a particular trajectory on the bloch sphere . by using msr , which represents a quantum pure state of spin - j systems in terms of a symmetrized state of 2j spin-1/2 systems
, one can generalize this geometric approach to large spin systems or multi - level systems . in terms of msr
, the evolution of a spin-@xmath0 state can be intuitively understood by trajectories of @xmath1 points on the two - dimensional ( 2d ) bloch sphere , with these @xmath2 points generally coined as majorana stars ( mss ) .
this approach naturally provides an intuitive way to study high spin systems from geometrical perspectives , which has made the msr a useful tool in many different fields , e.g. , classification of entanglement in symmetric quantum states @xcite , analyzing the spectrum of the lipkin - meshkov - glick model @xcite , studying bose condensate with high spins @xcite , and calculating geometrical phases of large - spin systems @xcite .
as much of the search for the application of msr has focused on high - spin systems , it is interesting to apply this approach to study the multi - band topological systems .
it is well known that a topological insulator distinguishes a trivial band insulator by its nontrivial topological energy band , which has different geometrical property from a trivial band @xcite .
for a two - band system , e.g. , the su - schrieffer - heeger ( ssh ) model @xcite , one can map the bloch state into a 2d bloch sphere , and the geometrical meaning of topologically different phases can be unveiled by their distinct trajectories @xcite . as a paradigmatic topological model @xcite , the ssh model supports either topologically trivial or nontrivial phase , characterized by the quantized berry phase @xmath3 or @xmath4 @xcite , which is experimentally measurable as demonstrated in the recent cold atom experiment @xcite . in this work
, we shall apply msr to study one - dimensional ( 1d ) topologically nontrivial polymerized systems , which can be viewed as a multi - band generalization of the celebrated dimerized ssh model @xcite and were also proposed to be realizable in optical superlattice systems @xcite .
the application of msr enables us to investigate the geometrical property of muti - band topological systems by mapping the multi - level quantum states to mss on the bloch sphere .
consequently , a filled bloch band forms specific trajectory of mss on the bloch sphere , which exhibits very different geometrical structure for topological or trivial phase .
furthermore , we unveil that the distinct geometrical structures of trajectories for topologically different states are closely related to their parities and determined by the distribution of mss at two high - symmetry momentum points .
our study provides an intuitive way to distinguish topologically different phases of multi - band systems and unveils the intrinsic relation between the band topology and parity from the geometrical point of view , which shall deepen our understanding of their geometrical properties .
we consider a 1d superlattice with period @xmath5 @xmath6 described by the hamiltonian : @xmath7 where @xmath8 and we require the system having inversion symmetry by enforcing @xmath9 for the case with @xmath10 , it reduces to the ssh model @xcite .
models with @xmath11 can be viewed as generalizations of the dimerized ssh model @xcite , and for convenience , we also refer them as polymerized models . as the unit cell consists of @xmath5 sites , the spectrum of the system splits into @xmath5 bands . by taking a fourier transformation @xmath12 with @xmath13 , the hamiltonian in the momentum space
can be written as @xmath14 with @xmath15 and @xmath16 for the general case with arbitrary @xmath17 , the system does not have topologically nontrivial properties .
nevertheless , when the system has inversion symmetry , it may support topologically nontrivial phase characterized by the quantized nontrivial berry phase or the emergence of doubly degenerate edge states for the open chain @xcite . for convenience
, we shall focus our study on a concrete case with @xmath18 in the present work , and also show our results can be directly generalized to cases with larger periods , e.g. , the case with @xmath19 . for the @xmath18 superlattice model or trimerized model with inversion symmetry
, the hopping amplitudes can be parameterized as @xmath20 where @xmath21 is taken as the unit of energy and @xmath22 is set .
the spectrum is split into @xmath23 bands and gap between bands is always open for any nonzero @xmath24 . considering the state with the lowest band being fully filled ,
we find that the state is topologically different for @xmath25 and @xmath26 . for the open chain with the length @xmath27
, there appear degenerate edge states at both ends for @xmath28 but none for @xmath29 .
this result suggests that there exists a topological phase transition by varying @xmath24 with the transition point at @xmath30 . for the periodic chain
, the topological phase transition can be characterized by the change of the berry phase of the system , i.e. , the berry phase @xmath31 in the topological phase and @xmath32 in the trivial phase . here the berry phase across the brillouin zone ( bz ) , also known as zak phase @xcite , is defined as @xmath33 with @xmath34 denoting the occupied bloch states .
the existence of topological states in our polymerized models is protected by inversion symmetry @xcite . in the momentum space
, the inversion symmetry means that @xmath35 , where the inversion operator @xmath36 is an anti - diagonal matrix with the the matrix element given by @xmath37 .
the system also has time reversal symmetry , which leads to @xmath38 , here the time reversal operator @xmath39 is just the complex - conjugation operator @xmath40 .
as we shall see in the following context , both the inversion and time reversal symmetries give some restrictions on the msr of bloch states , which plays an important role in determining trajectories of mss of topologically different states .
for a @xmath5-band system , the bloch state can be expressed as @xmath41 . to represent this multi - level state by mss ,
it is convenient to map the state to a spin - j state @xmath42 with @xmath43 .
there is a one - to - one correspondence between parameters @xmath44 and @xmath45 by taking @xmath46 . according to schwinger boson representation theory @xcite
, the angular momentum operators can be described by creation and annihilation operators of two mode bosons , @xmath47 , @xmath48 , and @xmath49 , @xmath50 , and the state @xmath51 can be expanded by @xmath52 , where @xmath53 is defined by @xmath54 and @xmath55 . with the help of schwinger representation
, the state @xmath34 can be factorized as @xmath56 where @xmath57 is the normalization coefficient .
if we denote @xmath58 , then the factorization parameters @xmath59 and @xmath60 can be determined by the roots of the following polynomial equation @xcite @xmath61 from eq.([phi ] ) , it is obvious that @xmath62 can be viewed as the product of @xmath2 spin-@xmath63 states with @xmath64 , and a given @xmath5-band bloch state can be described by @xmath65 mss on the bloch sphere . considering the time reversal symmetry and inversion symmetry of the inversion - invariant polymerized model , we can directly get the eigenstate @xmath66 fulfilling the following relations : @xmath67 ) , we get @xmath68 .
so if @xmath69 is the solution of eq.([poly ] ) , then @xmath70 must be the solution of @xmath71 in other words , for any star @xmath72 of @xmath73 , there must exist a star @xmath74 satisfied @xmath75 , @xmath76 .
so the whole trajectory of mss for a filled bloch band is symmetric about the meridian of the bloch sphere .
similarly , from eq.([eq2 ] ) , we have @xmath77 , @xmath78 .
so if @xmath79 is a solution of eq.([poly ] ) , @xmath80 is also a solution .
that is to say , for a fixed momentum k , any star @xmath81 of @xmath82 corresponds to a star @xmath83 of @xmath84 , unless @xmath81 lies on the equator .
so the whole trajectory of mss for a filled band is always symmetric about the equator . , the berry phase is @xmath85 , @xmath86 , @xmath87 .
( b ) topologically nontrivial state with @xmath88 , the berry phase is @xmath89 , @xmath90 , @xmath87 .
( c ) and ( d ) show the change of @xmath91 of mss for the lowest bloch band of trimerized model as @xmath92 goes from @xmath93 to @xmath4 , corresponding to ( a ) and ( b ) , respectively .
, title="fig : " ] + in terms of the msr , the berry phase for the @xmath5-band polymerized model can be represented as a sum of two parts , @xmath94 , where @xmath95 and the phase @xmath96 represents correlations between any two mss @xcite , where @xmath97 is a function of @xmath92 and the integral is carried out along the evolution paths of mss as the momentum goes over the bz . by using symmetries of mss
, we can prove that @xmath98 as the summation terms of correlations cancel out each other ( see the appendix for details ) .
consequently , the berry phase of the occupied bloch band is simplified to the sum of berry phases of mss on the bloch sphere , generated by the evolution of each star .
since the mss either appear in pairs , in terms of @xmath81 and @xmath99 for @xmath100 , or locate in the equator , we can further simplify the expression of berry phase by using @xmath101 .
finally , the berry phase of an occupied bloch band is simplified to @xmath102 which is the sum of integrals of each ms along its projecting trace on the equator . for the trimerized model parameterized by eq.([trimer ] )
, the eigenstate @xmath103 can be described by two mss . as @xmath92 goes across the bz ,
trajectories of the lowest bloch band for cases with @xmath26 and @xmath25 are shown in fig.[fig1](a ) and ( b ) , respectively .
it is obvious that trajectories for topologically different states display distinct geometric structures . here ,
in the topologically trivial phase , the berry phase of each star cancels out .
however , in the topologically nontrivial phase , the trajectories of two stars splice together to cover the whole equator , which gives @xmath104 . to see this more clearly
, we show the change of @xmath105 as a a function of momentum in fig.[fig1](c ) and ( d ) , corresponding to fig.[fig1](a ) and ( b ) , respectively . for the topologically trivial case ,
as @xmath92 goes from @xmath93 to @xmath4 , the integral over @xmath91 for each star cancels out each other , which generates a zero zak phase .
next we show that different topological structures of trajectories of mss are closely related to properties of mss at @xmath106 and @xmath107 . at these high - symmetry points denoted by @xmath108 , which fulfils @xmath109 up to a reciprocal lattice vector
, we have @xmath110 and @xmath111 .
these symmetries suggest that the distribution of mss for the high - symmetry state @xmath112 must be mirror - symmetrical about both the meridian and equator .
the state @xmath112 also has a certain parity , i.e. , @xmath113 with @xmath114 , and the parity @xmath115 of a filled bloch band is given by the product of @xmath116 and @xmath117 . using the expression of @xmath36
, we get @xmath118 where all coefficients @xmath119 are real due to the time - reversal symmetry . substituting ( [ app21 ] ) into the eq.(6 ) , we find that if @xmath120 is the solution of the equation , then @xmath121 , @xmath122 and @xmath123 are also the solutions , i.e. , the majorana stars for the high - symmetry state are mirror - symmetrical about both the meridian and equator , as schematically shown in fig .
[ pro2 ] . if @xmath124 , the four stars are degenerated to 2 stars on the equator ; if @xmath125 or @xmath4 , the stars are degenerated to two stars on the meridian ; and if @xmath124 and @xmath125 or @xmath4 , all the stars are degenerated to a single star . for the odd - parity state , @xmath126 , and we get @xmath127 from eq .
( [ app22 ] ) . while for the even - parity state , @xmath128 , and we get @xmath129
. + for the trimerized model with @xmath18 , there are only two majorana stars corresponding to a high - symmetry momentum state . for the odd - parity state
, the stars must be located at @xmath130 and @xmath131 in order to fulfil the mirror - symmetry conditions and the condition of @xmath132 simultaneously . while for the even - parity state
, the stars can be located at @xmath133 and @xmath134 or at @xmath135 and @xmath136 , or at @xmath137 and @xmath138 . to see it clearly
, we display the distribution of mss at the high - symmetry points for both the topologically trivial ( in fig.[pot](a ) ) and nontrivial phases ( in fig.[pot](b ) ) . while the parity of @xmath139 is always even with @xmath140 for both phases , we have @xmath141 for the trivial phase , and @xmath142 for the topological phase . and
( b ) @xmath88 , respectively .
blue points and red stars correspond to the state at @xmath106 and @xmath107 , respectively . while the parity for the point of @xmath107 is always even , the parity for the point of @xmath106 can be even for the trivial phase ( a ) or odd for the topological phase ( b ) .
, title="fig : " ] + goes from @xmath3 to @xmath4 , ( a ) for topologically trivial phase , ( b ) and ( c ) for topologically nontrivial phase . here
@xmath143 correspond to mss at @xmath106 and @xmath144 to mss at @xmath107.,title="fig : " ] + as the topologically trivial and nontrivial phases have different parities , next we unveil the berry phase of the corresponding bloch band can be uniquely determined by distributions of mss at two high - symmetry points . to calculate the berry phase ,
it is convenient to project the trajectory of mss on the bloch sphere to the equator , according to eq.([phase ] ) .
consequently the mss symmetrically located above and below the equator merge to the same point .
then , we divide the bz into two parts @xmath145 and @xmath146 , and we just need to consider the interval @xmath145 as the integral of @xmath91 over the interval @xmath146 ( equivalently @xmath147 ) gives the same contribution to the berry phase as over the interval @xmath145 , according to the symmetry analysis . hence , the zak phase @xmath148 is just the double of the berry phase as @xmath92 integrates from @xmath3 to @xmath4 . in the topologically trivial phase , when @xmath106 , both mss are located at @xmath149 " on the projected equator shown in fig.[exp](a ) .
as @xmath92 travels from @xmath3 to @xmath4 , the stars move to @xmath150 and @xmath151 , generating changes of polar angles @xmath152 and @xmath153 , respectively , where @xmath154 can be any integer . from eq.([phase ] ) , we get the zak phase @xmath155 , i.e. , @xmath156 . in the topologically nontrivial phase , mss corresponding to the @xmath106 state
are located at @xmath149 " and @xmath157 " , and there are two possibilities for the evolution of state from @xmath106 to @xmath107 , as shown in fig.[exp](b ) and ( c ) . for the case of fig.[exp](b ) , @xmath158 moves to @xmath150 and @xmath159 to @xmath151 as @xmath92 goes from @xmath3 to @xmath4 , with corresponding changes of polar angles given by @xmath160 and @xmath161 .
since @xmath162 , we get the zak phase @xmath163 , i.e. , @xmath164 . for the case of fig.[exp](c ) , we can get the same conclusion by following similar discussions .
our results can be directly generalized to general polymerized models with inversion symmetry .
concretely , we consider a tetramerized model with @xmath165 , @xmath166 and @xmath167 , and take @xmath21 and @xmath22 .
similar to the case of @xmath18 , there exists a topological phase transition occurring at @xmath30 , with @xmath26 corresponding to the trivial phase and @xmath25 the topologically nontrivial phase .
the analysis of geometric meaning is analogous to the trimerized model despite a little more complicated . and
( b ) @xmath88 , respectively .
the blue , red and green curves represent different mss .
blue points and red stars represent mss of states at @xmath106 and @xmath107 , respectively .
( c ) and ( d ) display the change of @xmath91 of mss as @xmath92 goes from @xmath93 to @xmath4 , corresponding to ( a ) and ( b ) , respectively .
, title="fig : " ] + for the tetramerized model with @xmath19 , each state in the momentum space is represented by three mss , and their trajectories over the bz are shown in fig.[fig4](a ) and ( b ) for topologically trivial and nontrivial phases , respectively .
mss in two high - symmetry points are also marked on the trajectories . in the topologically trivial phase , all trajectories of mss
go back and forth , whereas in the topologically nontrivial phase , the trajectories form a close circle and cover the equator completely . consequently ,
zak phases for the topologically trivial and nontrivial states are @xmath3 and @xmath4 , corresponding to fig.[fig4](c ) and ( d ) , respectively .
similarly , different ms distributions at @xmath107 for topologically trivial and nontrivial states indicate that they have different parities .
in summary , we have unveiled the geometrical properties of topological phases of 1d inversion - invariant multi - band systems from trajectories of mss , which exhibit different topological structures for topologically different phases . by utilizing the symmetric properties of mss
, we found that the zak phase of a filled band can be represented as the summation of integral of each ms along its projecting trace on the equator , and takes @xmath3 and @xmath4 for topologically trivial and nontrivial phases , respectively .
we further demonstrated that the topological structure of the trajectory of mss is closely related to the parity of the system , which is determined by properties of bloch states at two high - symmetry points .
s. c. is supported by nsfc under grants no .
11425419 , no .
11374354 and no . 11174360 and by national program for basic research of most .
h. g. is supported by nsfc under grants no.11274032 and no .
. l. b. f. is supported by the nsfc under grants no .
the correlation phase @xmath168 in the main text can be represented as @xmath169 with @xmath170 where the normalization coefficient @xmath57 is a symmetric function for permuting any two stars @xmath97 and @xmath81 , and @xmath171 is the distance .
it is clear that @xmath168 is a sum of correlations between any two stars . due to the existence of time reversal symmetry and inversion symmetry , the majorana stars for a bloch state with fixed momentum @xmath92 distribute either on the equator or symmetrically about the equator .
so the distribution of majorana stars for a bloch state can be schematically displayed in fig .
[ sfig1 ] , where @xmath172 are the stars above the equator , @xmath173 are the corresponding ones below the equator symmetrical to @xmath172 , and @xmath174 are on the equator .
firstly , we calculate the correlation between @xmath172 and @xmath175 and the corresponding correlation between @xmath173 and @xmath176 , which are given by @xmath177 and @xmath178 where @xmath179 since @xmath180 and @xmath181 , we have @xmath182 , @xmath183 , @xmath184 and @xmath185 , where @xmath186 and @xmath187 . substituting them into ( [ p1 ] ) and ( [ p2 ] ) we find that @xmath188 .
similarly , we can derive @xmath189 , @xmath190 .
for the correlation between @xmath172 and @xmath173 , we know that @xmath191 and @xmath192 , so @xmath193 at last , the correlation between @xmath194 and @xmath195 is always @xmath3 , because they are on the same plane so that @xmath196 vanishes apparently .
hence , we have @xmath197 and the sum of correlation phases @xmath198 is zero as the summation terms cancel out each other . | by using majorana s stellar representation , we give a clear geometrical interpretation of the topological phases of inversion - symmetric polymerized models by mapping the bloch states of multi - band systems to majorana stars on the bloch sphere . while trajectories of majorana stars of a filled bloch band exhibit quite different geometrical structures for topologically trivial and nontrivial phases , we further demonstrate that these structures are uniquely determined by distributions of majorana stars of two high - symmetrical momentum states , which have different parities for topologically different states . |
energy is fundamental to science in general and the life sciences in particular @xcite . for this reason , the energy - based bond graph modelling method , originally developed in the context of engineering @xcite ,
has been applied to modelling biomolecular systems @xcite .
bond graphs represent the energy consumption per unit time , or power flow , for each component of a given system .
power flows are calculated as the product of appropriately chosen ` signal quantities ' named ` efforts ' and ` flows ' .
examples of effort include force , voltage , pressure and chemical potential . corresponding examples of flow are velocity , current , fluid flow rate and molar flow rate . by describing systems under the unifying principle of energy flows and due to the abstract representation of these flows in terms of generalised effort and flow quantities , the bond graph approach is particularly appropriate for modelling systems with multiple energy domains .
a comprehensive account of the use of bond graphs to model engineering systems is given in the textbooks of @xcite , @xcite , @xcite and @xcite and a tutorial introduction for control engineers is given by @xcite .
chemical reactions are considered by @xcite and @xcite . in particular , as an engineering example , @xcite has looked at chemoelectrical energy flows in hybrid vehicles .
chemoelectrical energy transduction is fundamental to living systems and occurs in a number of contexts including oxidative phosphorylation and chemiosmosis , synaptic transmission and action potentials in excitable membranes .
for this reason , this paper extends the energy - based bond graph modelling of biomolecular systems to biological systems which involve chemoelectrical energy transduction .
a simple model of the action potential in excitable membranes is used as an illustrative example of the general approach .
although such simple models could easily be derived by other means , our general approach could be used to build thermodynamically compliant models of large hierarchical systems such as those describing metabolism , signalling and neural transmission .
understanding the biophysical processes which underlie the generation of the action potential in excitable cells is of considerable interest , and has been the subject of intensive mathematical and computational modelling . since the early work of @xcite on modelling the ionic mechanisms which give rise to the action potential in neurons , mathematical models of the action potential
have incorporated ever - increasing biophysical and ionic detail @xcite , and have been formulated to describe both normal and pathophysiological mechanisms .
generation of the action potential comes at a metabolic cost .
energy is required to maintain the imbalance of ionic species across the membrane , such that when ion channels open there is a flux of ions ( current ) across the membrane initially carried by sodium ions generating rapid membrane depolarisation ( the upstroke of the action potential ) .
each action potential reduces the ionic imbalance and each ionic species needs to be transported across the membrane against an adverse electrochemical gradient to restore the imbalance this requires energy .
the role of energy in neural systems has been widely discussed in the literature @xcite and it has been suggested that metabolic cost is a unifying principle underlying the functional organisation and biophysical properties of neurons @xcite .
furthermore , @xcite posed the question `` does impairment of energy metabolism result in excitotoxic neuronal death in neurodegenerative illnesses ? '' more recently , it has been suggested @xcite that an energy - based approach is required to elucidate neuro - degenerative diseases such as parkinson s disease . in such studies , the flow of @xmath0 across the membrane
is taken as a proxy for energy consumption associated with action potential generation , as @xmath0 has to be pumped back across the membrane by an energy - consuming atpase reaction .
this energetic cost is often quoted as an equivalent number of atp molecules required to restore the ionic concentration gradient through activity of the sodium - potassium atpase ( the na pump ) , as calculated via stoichiometric arguments . while this provides a useful indication of energetic cost , this is however an imprecise approach , which can not produce reliable estimates of energy flows under all conditions ( physiological and pathophysiological ) .
what is required instead is a way of simulating and calculating the actual energy flows associated with these ionic movements through a physically - based modelling approach . here
we develop a general bond graph based modelling framework that enables explicit calculation of the energy flows involved in moving ions across the cell membrane .
our model can be applied to the regulation of the membrane potential via ion channels in any cell type
obvious examples include , for example , cardiac and skeletal muscle cells but in order to give a specific application we investigate the energy cost for generating a neural action potential . because ions carry electrical charge , differences in the concentration of a particular ionic species on either side of a membrane generate both a chemical potential difference due to the concentration gradient as well as an electrical potential difference due to charge imbalance .
we derive a bond graph formulation for the voltage - dependent ionic flow across a membrane that comprises both the classical hodgkin - huxley model and its more detailed variants focussing on the associated flow of energy . one component of this formulation accounts for the behaviour of the ion channel as an electrical resistor that can be investigated experimentally via current - voltage relationships , also known as @xmath1-@xmath2 curves .
popular models for @xmath1-@xmath2 curves are , for example , the linear @xmath1-@xmath2 dependency that corresponds to ohm s law and the non - linear goldman - hodgkin - katz ( ghk ) equations .
hodgkin - huxley - like models assume a linear @xmath1-@xmath2 relationship .
@xcite represented the voltage - dependent opening and closing of ion channels by an empirical model , their gating variables , which led to excellent agreement with experimental data .
this paper replaces this empirical model of gating by an physically - based voltage - dependent markov model with one open and one closed state .
whereas this model accounts for the energy needed for moving the ion channel between conformations where the channel is in an open or a closed state , respectively , the empirical gating variables of the hodgkin - huxley model do not have a similar interpretation . as our models are formulated using bond graphs , important physical quantities such as mass and energy
are conserved ( in the sense that any dissipative processes are directly accounted for , and any inputs to or losses from the system are quantified ) .
this theoretical approach is then applied to analyse energy consumption in retinal ganglion cells based on _ in vitro _ experimental data collected and analysed from retinal ganglion cells ( rgcs ) of wild - type ( wt ) and degenerative ( rd1 ) mice .
our use of the rd1 degenerate retina mouse model ensures that the outcomes of this project are directly relevant to human patients since rd1 mice have a degenerate retina that has distinct similarities to that observed in human patients with _
retinitis pigmentosa _
a set of hereditary retinal diseases that results from the degenerative loss of the photoreceptors in the retina .
a virtual reference environment @xcite is available for this paper at https://github.com/uomsystemsbiology/energetic_cost_reference_environment .
external to the membrane is represented by @xmath3:@xmath4 , the ion @xmath5 internal to the membrane is represented by @xmath3:@xmath6 and the internal to external molar flow rate is @xmath7 .
@xmath8:@xmath9 is the electrostoichiometric transformer where @xmath10 is the integer ionic charge .
@xmath3:@xmath11 represents the membrane potential as an electrogenic capacitance with electrogenic potential @xmath12 and electrogenic flow @xmath13 .
the ionic flow is determined by the reaction component @xmath14:@xmath15 and modulated by the gating affinity @xmath16 associated with the @xmath3:@xmath17 .
the two components @xmath3:@xmath17 and @xmath3:@xmath11 and associated junctions may be relaced by ports ; this enables the model to be reused within a hierarchical framework .
, title="fig : " ] [ subfig : channelreac_abg ] we represent the flow of ions through the open pore of an ion channel in analogy to a chemical reaction : @xmath18 { } } } x_e + g\ ] ] the intracellular and extracellular concentrations of a particular ion are represented as different chemical species , the intracellular species @xmath5 and the extracellular species @xmath19 .
conversion from @xmath5 to @xmath19 and vice versa is modelled by an enzymatic reaction with the _ gating species _ @xmath20 that `` catalyses '' the flow of ions across the cell membrane and accounts for the voltage - dependent opening and closing of the channel as well as the behaviour of the channel as an electrical resistor .
the influence of the membrane potential is represented in this reaction by the _
electrogenic species _ @xmath21 . further below we
will explain how an electrical potential @xmath2 can be converted to an equivalent chemical potential @xmath22 .
a bond graph representation of according to the framework developed by @xcite is given in figure [ fig : bg_channel ] .
using this specific example we briefly review the bond graph approach developed in our previous work .
the components of a bond graph are distinguished based on how they transform energy .
capacitors or springs store energy ( ) , resistors or dampers dissipate energy ( ) , and transducers ( or transformers ) ( ) which transmit and convert , but do not dissipate , power .
the chemical species appearing in are represented in the bond graph by capacitors @xmath3:@xmath6 , @xmath3:@xmath4 , @xmath3:@xmath11 and @xmath3:@xmath17 . because a given type of component usually occurs more than once in a given system , the ` colon ' notation is adopted to distinguish between different instances of each component type : the symbol preceding the colon identifies the type of component , and the label following the colon identifies the particular instance .
let us first consider the species @xmath23 and @xmath24 .
these are simply the concentrations of a particular ion within the cell ( @xmath23 ) and outside the cell ( @xmath24 ) so that their rates of change must equal the molar flow rate @xmath7 in with opposite sign : @xmath25 it is assumed that in the body biochemical reactions occur under conditions of constant pressure ( isobaric ) and constant temperature ( isothermal ) . under these conditions ,
the chemical potential @xmath26 of substance @xmath27 measured in is given @xcite in terms of its mole fraction @xmath28 as : @xmath29 where @xmath30 is the value of @xmath26 when @xmath27 is pure ( @xmath31 ) , @xmath32 is the universal gas constant , @xmath33 is the absolute temperature and @xmath34 is the natural ( or napierian ) logarithm . by introducing the _ thermodynamic constant _
@xmath36 where @xmath37 is the total number of moles in the mixture we can express the chemical potential as a function of molar amount @xmath38 of the species @xmath27 : @xmath39 as discussed by @xcite the chemical potential @xmath12 and the molar flow @xmath13 are appropriate _ effort _ and _ flow _ variables for modelling chemical reactions .
the product of chemical potential @xmath12 and molar flow @xmath13 is the energy flow into the bond graph components and has the unit of _ power_.
this is shown in figure [ fig : bg_channel ] by power bonds ( or more simply ` bonds ' ) , drawn as harpoons : @xmath40 .
these bonds can optionally be annotated with specific effort and flow variables , for example @xmath41{e}$ ] .
the half - arrow on the bond indicates the direction in which power power flow will be regarded as positive and thus defines a sign convention . as explained above , for the components @xmath3:@xmath6 and @xmath3:@xmath4 we have the flows @xmath42 and @xmath7 , respectively , and the chemical potentials are @xcite : @xmath43 @xmath44 is the standard chemical potential for the ion and the volumetric capacities of the interior and exterior are @xmath45 and @xmath46 , respectively . in general ,
bond graph components which are connected by a bond share the same effort and flow .
thus , because all energy flows between bond graph components are quantified , the bond graph ensures that properties such as energy and mass are conserved .
we follow @xcite in describing chemical reactions in terms of the _ marcelin de donder _ formulae as discussed by @xcite and @xcite . in particular , given the @xmath47th reaction ( * ? ? ?
* ( 5.9 ) ) : @xmath48{i } } } \nu^r_a a + \nu^r_b b + \nu^r_c c \dots\ ] ] where the stoichiometric coefficients @xmath49 are either zero or positive integers and the _ forward affinity _ @xmath50 and the _ reverse affinity _
@xmath51 are defined as : @xmath52 thus , the affinities @xmath50 and @xmath51 are sums of the chemical potentials appearing on either side of weighted with the stoichiometric coefficients @xmath49 . in a bond graph model , the affinities from , can be obtained via a ` junction ' .
junctions allow parallel ( ` ' junction ) and series ( ` ' junction ) connections to be made .
the efforts on bonds impinging on a junction sum to zero whereas the flows are all equal ( this is equivalent to kirchhoff s second law for electrical circuits , where effort is equivalent to voltage and flow is equivalent to current ) .
analogously , for a junction the flows sum to zero and the efforts are all equal ( equivalent to kirchhoff s first law ) .
taking into account the sign convention explained above it is easy to see that @xmath53 and @xmath54 in figure [ fig : bg_channel ] are indeed weighted sums of chemical potentials as in , : @xmath55 the affinity of the gating species @xmath56 as well as the potential @xmath22 of the electrogenic species will be derived further below . the marcellin
de donder formulae enable us to calculate the flow of a reaction for which we assume the law of mass action from the affinities : @xmath57 note that the arguments of the exponential terms are dimensionless as are @xmath58 and @xmath59 .
the units of the reaction rate constant @xmath60 are those of molar flow rate : @xmath61 .
the flow @xmath62 is obtained from the efforts @xmath50 and @xmath51 , similar to an electrical resistor where ohm s law requires that the current is proportional to the potential difference across the resistor .
in contrast to the linear electrical component , the non - linear effort - flow relationship of can be represented by a two port resistive component , the component @xcite .
resistive components such as and components dissipate energy .
so far we have not accounted for the interdependency of the molar flow @xmath7 and the electrical potential @xmath2 across the cell membrane . as discussed in bond graphs terms by @xcite
the well - known physical relationship between the molar flow of ions @xmath7 across the membrane and the current @xmath47 @xmath63 in fact describes the coupling of two energy domains , namely chemical energy and electrical energy . here
@xmath10 is the ( integer ) ionic charge and @xmath64 faraday s constant which has the approximate value @xmath65 thus , for example , an ion with unit charge and flow rate of @xmath66 is equivalent to a current of about @xmath67 . into electrical current @xmath47
is represented by @xmath8:@xmath68 where where @xmath10 is the ionic charge and @xmath64 faraday s constant ; @xmath3:@xmath69 represents the membrane capacitance containing electrical charge @xmath70 and @xmath22 is the _ chemoelectrical potential _ : the chemical potential corresponding to the membrane voltage @xmath2 ( b ) @xmath8:@xmath68 is split into two series components @xmath8:@xmath9 & @xmath8:@xmath71 .
( c ) the electrical capacitor @xmath3:@xmath69 and the chemoelectrical transducer @xmath8:@xmath71 are combined into chemo - equivalent capacitor @xmath3:@xmath11 containing @xmath72 moles of charge .
@xmath3:@xmath11 is analogous to a chemical species , and @xmath8:@xmath9 to a stoichiometric transformer .
, title="fig : " ] [ subfig : cequiv_abg ] using the bond graph _
element denoted by , equation enables us to associate a _
chemoelectrical potential _
@xmath22 with an ionic current @xmath47 at a membrane potential @xmath2 .
as shown in figure [ fig : cequiv](a ) , the transformer @xmath8:@xmath68 converts electrical power @xmath73 to chemical power @xmath74 so that from @xmath75 we can derive the defining equation of the chemoelectrical potential @xmath76 thus , for example , a voltage of @xmath77 is equivalent to a chemical potential of an ion with unit charge of about @xmath78 stoichiometric analysis of biochemical reactions has a bond graph interpretation in terms of the bond graph structure . as a structural matrix
, the stoichiometric matrix for biochemical systems contains integer entries corresponding to reaction stoichiometry .
the bond graph describing chemoelectrical transduction in figure [ fig : cequiv](a ) contains transformers connecting the electrical and chemical domains and the chemoelectrical transformer contains the faraday constant @xmath64 which is not an integer . for this reason ,
it is useful as shown in figure [ fig : cequiv](b ) to split the component @xmath8:@xmath68 into two components in series : @xmath8:@xmath9 and @xmath8:@xmath71 .
thus @xmath8:@xmath9 corresponds to the integer , but ion - dependent charge , @xmath10 , and @xmath8:@xmath71 to the universal constant , but non - integer , @xmath64 .
@xmath8:@xmath9 will be referred to as the _ electrostoichiometric transformer _ with ratio @xmath10 . with reference to figures [ fig : cequiv](b)&(c ) 2 v & = zv_ion & _ v & = z_q furthermore , the electrical component @xmath3:@xmath69 and the chemoelectrical transformer @xmath8:@xmath71 component may be replaced by a single _ electrogenic capacitor _ @xmath3:@xmath11 as in figure [ fig : cequiv](c ) .
assuming that the electrical capacitor has the constant capacitance @xmath79 , and with reference to figure [ fig : cequiv](b ) 4[eq : c_actual ] _ q & = f v_q & v_q & = & & = i_q & i_q & = fv_p while the electrogenic capacitor @xmath3:@xmath11 is linear and therefore can not be written in the logarithmic form of equation , it is convenient to write the defining equations in a similar form by defining the number of moles of charge @xmath72 as @xmath80 with reference to figure [ fig : cequiv](c ) 2[eq : c_equiv ] _ p & = rt k_e x_m & _ m & = v_p comparing equations and , it follows that : @xmath81 @xmath82 is the _ equivalent voltage _ associated with each mole of charge and is dependent on the electrical capacitance @xmath83 .
@xmath84 is related to the _ nernst potential _ and is temperature dependent . in summary ,
@xmath3:@xmath11 is analogous to a chemical species ( but with a non - logarithmic characteristic ) , and @xmath8:@xmath9 is analogous to the stoichiometric transformers discussed by @xcite , @xcite and @xcite .
[ subfig : cc_abg ] because the influence of the membrane potential @xmath2 can be translated to the contribution of the electrogenic species , its effect on the flow of a species @xmath27 can be accounted for in the bond graph just like a reaction of two normal species by coupling the efforts @xmath22 and @xmath26 via a junction @xcite as shown in figure [ fig : cc ] . from the bond graph representation we rederive the standard electrostatic contribution to the chemoelectrical potential : @xmath85 in general ,
as discussed below , the electrogenic capacitor will correspond to the net flow of more than one charged species .
similar to our approach to modelling the membrane potential we will also represent the voltage - dependent gating of ion channels as a chemical species . in this way we obtain a general model for ionic flow through ion channels across cell membranes .
we will demonstrate that our general model of voltage - dependent gating includes the original hodgkin - huxley model and its extensions beyond sodium @xmath86 , potassium @xmath87 and leak currents .
as a specific application of our framework we will investigate the energy flow related to the action potential predicted by hodgkin - huxley like models .
our analysis clarifies that the phenomenological gating variables introduced by @xcite fail to account appropriately for energy flows related to ion channel gating . as shown in the chemical reaction that our model is based upon , the _ gating species _
@xmath20 behaves like an enzyme that catalyses the conversion of the intracellular species @xmath5 to the extracellular species @xmath19 . in the same way as in (
* figure 2e ) the _ gating affinity _
@xmath16 is added to both sides of the reaction ( figure [ fig : bg_channel ] ) .
the second port @xmath88:@xmath89}$ ] imposes the gating affinity @xmath16 and the corresponding flow @xmath90 . with the gating affinity @xmath91
we represent two characteristics of the ionic flow through a channel .
the term @xmath92 accounts for the fact that an ion channel has an electrical resistance that opposes the ionic current .
the electrical resistance is commonly investigated experimentally by determining current - voltage relationships .
it will be demonstrated that the two most commonly used models for current - voltage relationships of ion channels , ohm s law and the goldman - hodgkin - katz ( ghk ) equations , can be obtained by suitable choices of @xmath92 . whereas @xmath92 represents the conductance through an open channel
, @xmath93 provides a model for the voltage - dependent opening and closing of the channel . from equations ( [ eq : a^f]-[eq : v_exp ] ) we obtain our model for the ionic flow through an ion channel : @xmath94 \\ & = \kappa k_{ion } \exp{a^g(v ) } \left ( c_i \exp { \bar{v}}- c_e \right ) \label{eq : v_1}\\ \text{where } c_i & = \frac{x_i}{c_i},\ ; c_e = \frac{x_e}{c_e } , \ ; v_n=\frac{r t}{f } \text { and } { \bar{v}}:=\frac{v}{v_n } \end{aligned}\]]define @xmath95 as the voltage for which @xmath7 of equation ( [ eq : v_1 ] ) is zero : @xmath96 using equations ( [ eq : vv_0 ] ) and , equation becomes : @xmath97 where @xmath98 is given by @xmath99 the quantity @xmath100 is known as the _ ussing flux ratio _ @xcite .
the ionic flow ( ion current ) model of equation will be used in the sequel to construct a bond graph model from the hodgkin huxley model .
@xcite model an ion channel according to ohm s law i.e. as a linear conductance @xmath101 , modulated by a function @xmath102 in series with the nernst potential represented by a voltage source @xmath103 ( see figure [ fig : hh ] ) .
[ subfig : hh ] @xmath104 where @xmath105 is the gating function , which is a dynamic function of membrane potential @xmath2 . in terms of ionic flow @xmath7
, equation becomes , using : @xmath106 it is easy to see that our model contains the hodgkin - huxley model by a suitable choice of @xmath107 : @xmath108clearly , equation ( [ eq : g_hh ] ) only makes sense if @xmath109 is positive for all @xmath110 .
if @xmath111 , both @xmath110 and @xmath112 are negative ; if @xmath113 , both @xmath110 and @xmath112 are positive ; and , as @xmath114 , @xmath115 is positive for for all @xmath110 .
a number of alternative physically - based models for the ion channel are available .
in particular , the goldman - hodgkin - katz ( ghk ) model ( see ( * ? ? ?
* ( 2.123 ) ) , @xcite ) & @xcite can be rewritten in a similar form to ( [ eq : v_mass - action ] ) as : @xmath116 comparing equations ( [ eq : v_mass - action ] ) and ( [ eq : ghk ] ) , it follows that the mass action model and ghk model are the same if the model - dependent function @xmath117 is : @xmath118 note that @xmath119 is of the same form as @xmath120 except that @xmath110 is replaced by @xmath121 . from equations and ,
both the hh and ghk ion channel models give zero ionic flow when the membrane voltage equals the nernst voltage : that is the models match at @xmath122 .
moreover , the ghk model of equation has a parameter @xmath123 that can be chosen to fit the data . in this case , @xmath123 is chosen so that the ghk and hh models also match at another voltage ; in this case chosen as minus the nernst voltage .
figure [ fig : comparison ] shows the ionic currents plotted against membrane voltage for each of the three channels and they match at the two voltages .
the ghk model is used in the sequel . .ghk
parameter @xmath124 [ cols="<,<,<,<",options="header " , ] we have extended our earlier work on energy - based bond graph modelling of biochemical systems to encompass chemoelectrical systems . in particular , we have introduced the electrogenic capacitor and the electrostoichiometric transformer to bridge the chemical and electrical domains . as a particular example illustrating the general approach , we have constructed a bond graph model of the @xcite model of the axon ; and we have used this model to show that calculation of energy consumption during generation of the action potential by counting @xmath0 ions crossing the membrane underestimates true energy consumption by around 20% . in this particular situation , the concentrations are constant and thus equation is appropriate .
moreover , the contribution of the leakage and the gating currents are small and can be neglected .
the values of table [ tab : conc ] and equation gives the molar free energy values of table [ tab : res ] for @xmath125 , @xmath0 ; that for atp is taken from ( * ? ? ?
* ( 1.23 ) ) .
because the actual energy consumption depends on both the amounts of @xmath125 , @xmath0 as well as on the internal and external concentrations , there is no way that the atp - proxy formula ( based on only the amount of @xmath0 ) can give the correct value under all circumstances . to illustrate this , figure [ fig :
vary ] shows how actual and atp - proxy energy varies with internal @xmath0 concentration expressed as a ratio @xmath126 to the nominal concentration of table [ tab : conc ] .
the discrepancy between actual and atp - proxy energy varies with @xmath126 .
moreover , the method of this paper does _ not _ require the concentrations to be constant during an action potential and is thus applicable to more general situations .
the wider significance of our approach is that it provides a framework within which biophysically based models are robustly thermodynamically compliant @xcite , as required for example when considering the energetic costs and consequences of cellular biological processes .
furthermore , the bond graph approach provides a basis for modular modelling of large , multi - domain electro - chemical biological systems , such as is now commonplace in systems biology models of excitable membranes in the neuronal and cardiac contexts .
components and modules which are represented as bond graphs are physically plausible models which obey the basic principles of thermodynamics , and therefore larger models constructed from such modules will also consequentially be physically plausible models .
future work will further develop these concepts in order to represent ligand - gated ion channels , ion pumps ( such as the @xmath127 pump serca @xcite and the @xmath0 pump @xcite , as are required for current generation neuronal and cardiac cell models .
this modular approach allows simpler modules to be replaced by more complex modules , or empirical modules to be replaced by physically - based modules , as the underlying science advances .
furthermore , the multi - domain nature of bond graphs makes possible extension of the approach to mechano - chemical transduction . in actively contracting cardiac muscle , for example , energetic considerations are dominated by force production , where approximately 7580% of atp consumption in cardiomyocytes over a cardiac cycle is due to formation of contractile cross - bridges , 510% due to the @xmath0 pump , and @xmath127 extrusion and uptake into @xmath127 stores accounting for the remainder @xcite . in cardiac muscle
, energetics is known to play a critical role in the health of cardiac muscle , with many studies implicating energetic imbalance or inadequacy of energy production in heart disease @xcite .
models which provide a mechanism with which to assess the energetic aspects of cell function are therefore much needed . combining metabolism ,
electro - chemical and chemo - mechanical energy transduction to examine energy flows within the heart @xcite is therefore a major goal of our work .
conservation of mass and energy have been used by @xcite and @xcite to examine the dynamics of seizure and spreading depression .
it would be interesting to reexamine this work in the more general context of bond graph modelling .
the atp - proxy approach is based on assuming that the biological entity is operating in a normal state and therefore could lead to misleading conclusions in a pathophysiological state .
in contrast , our approach makes no assumption of normality and may be expected to be of use in pathophysiological states in general and , in particular , the retinal example discussed in this paper .
our use of the rd1 degenerate retina mouse model ensures that the outcomes of this project are directly relevant to human patients since rd1 mice have a degenerate retina that has distinct similarities to that observed in human patients with _
retinitis pigmentosa _ a set of hereditary retinal diseases that results from the degenerative loss of the photoreceptors in the retina .
it has been proposed that the death of rod photoreceptors results in decreased oxygen consumption @xcite .
in addition , it has been shown that potassium channel - opening agents directly affect mitochondria @xcite .
therefore , it is important to understand how energy consumption in degenerative retina is altered . the proposed methodology allows a comparison between the energy consumption in healthy and degenerate mice , even when the differences in action potentials between the two types are small .
the modularity of the bond graph approach allows the action potential models of this paper to be combined with models of the various trans - membrane pumps and transporters to give a thermodynamically correct model of atp consumption .
it would therefore be interesting to reexamine the optimality arguments of @xcite and @xcite using this approach .
future work will combine the chemoelectrical bond graph models of this paper with the bond graph models of anaerobic metabolism developed by @xcite , and bond graph models of aerobic metabolism currently under development , to give integrated models of neuronal energy transduction suitable for investigating neuronal dysfunctions such as parkinson s disease @xcite .
peter gawthrop would like to thank the melbourne school of engineering for its support via a professorial fellowship .
this research was in part conducted and funded by the australian research council centre of excellence in convergent bio - nano science and technology ( project number ce140100036 ) .
the theory was developed by pjg , is and ec .
the theoretical part of the paper was written by pjg , is and ec and the experimental part by tk .
experiments were conducted by ss and mi .
a virtual reference environment @xcite is available for this paper at https://github.com/uomsystemsbiology/energetic_cost_reference_environment .
45 natexlab#1#1url # 1`#1`urlprefix atkins , p. , de paula , j. , 2011 .
physical chemistry for the life sciences , 2nd edition .
oxford university press .
beal , m. f. , 1992 .
does impairment of energy metabolism result in excitotoxic neuronal death in neurodegenerative illnesses ?
annals of neurology 31 ( 2 ) , 119130 .
borutzky , w. , 2011 .
bond graph modelling of engineering systems : theory , applications and software support .
springer .
carter , b. c. , bean , b. p. , 2009 .
sodium entry during action potentials of mammalian neurons : incomplete inactivation and reduced metabolic efficiency in fast - spiking neurons .
neuron 64 ( 6 ) , 898 909 .
cellier , f. e. , 1991 . continuous system modelling .
springer - verlag .
cloutier , m. , bolger , fiachra , b. , lowry , john , p. , wellstead , p. , 2009 .
an integrative dynamic model of brain energy metabolism using in vivo neurochemical measurements .
journal of computational neuroscience 27 ( 3 ) , 391414 .
cloutier , m. , middleton , r. , wellstead , p. , june 2012 .
feedback motif for the pathogenesis of parkinson s disease .
systems biology , iet 6 ( 3 ) , 8693 . gawthrop , p. j. , bevan , g. p. , april 2007 .
bond - graph modeling : a tutorial introduction for control engineers .
ieee control systems magazine 27 ( 2 ) , 2445 .
gawthrop , p. j. , crampin , e. j. , 2014 .
energy - based analysis of biochemical cycles using bond graphs .
proceedings of the royal society a : mathematical , physical and engineering science 470 ( 2171 ) , 125 , available at arxiv:1406.2447 .
gawthrop , p. j. , crampin , e. j. , march 2016 .
modular bond - graph modelling and analysis of biomolecular systems .
iet systems biology 10 , available at arxiv:1511.06482 .
gawthrop , p. j. , cursons , j. , crampin , e. j. , 2015 .
hierarchical bond graph modelling of biochemical networks .
proceedings of the royal society a : mathematical , physical and engineering sciences 471 ( 2184 ) , 123 , available at arxiv:1503.01814 .
gawthrop , p. j. , smith , l. p. s. , 1996 .
metamodelling : bond graphs and dynamic systems .
prentice hall , hemel hempstead , herts , england .
greifeneder , j. , cellier , f. , 2012 . modeling chemical reactions using bond graphs . in : proceedings
icbgm12 , 10th scs intl . conf . on bond graph modeling and simulation .
genoa , italy , pp .
110121 .
hasenstaub , a. , otte , s. , callaway , e. , sejnowski , t. j. , 2010 .
metabolic cost as a unifying principle governing neuronal biophysics .
proceedings of the national academy of sciences 107 ( 27 ) , 1232912334 .
hille , b. , 2001 .
ion channels of excitable membranes , 3rd edition .
sinauer associates , sunderland , ma , usa .
hodgkin , a. l. , huxley , a. f. , 1952 . a quantitative description of membrane current and its application to conduction and excitation in nerve . the journal of physiology 117 ( 4 ) , 500544 .
hurley , d. g. , budden , d. m. , crampin , e. j. , 2014 .
virtual reference environments : a simple way to make research reproducible .
briefings in bioinformatics .
kameneva , t. , meffin , h. , burkitt , a. , 2011 .
modelling intrinsic electrophysiological properties of on and off retinal ganglion cells .
journal of computational neuroscience 31 ( 3 ) , 547561 . karnopp , d. , 1990 .
bond graph models for electrochemical energy storage : electrical , chemical and thermal effects .
journal of the franklin institute 327 ( 6 ) , 983 992 .
karnopp , d. c. , margolis , d. l. , rosenberg , r. c. , 2012 .
system dynamics : modeling , simulation , and control of mechatronic systems , 5th edition .
john wiley & sons .
katz , a. m. , 2011 .
physiology of the heart , 5th edition .
lippincott williams and wilkins , philadelphia .
keener , j. p. , sneyd , j. , 2009 . mathematical physiology : i : cellular physiology , 2nd edition .
vol . 1 .
springer .
koch , c. , 2004 .
biophysics of computation : information processing in single neurons .
oxford university press , oxford .
kulawiak , b. , kudin , a. p. , szewczyk , a. , kunz , w. s. , 2008 .
bk channel openers inhibit ros production of isolated rat brain mitochondria . experimental neurology 212 ( 2 ) , 543 547 .
mukherjee , a. , karmaker , r. , samantaray , a. k. , 2006 .
bond graph in modeling , simulation and fault indentification . i.k .
international , new delhi , .
neubauer , s. , 2007 .
the failing heart
an engine out of fuel .
new england journal of medicine 356 ( 11 ) , 11401151 .
niven , j. e. , laughlin , s. b. , 2008 .
energy limitation as a selective pressure on the evolution of sensory systems .
journal of experimental biology 211 ( 11 ) , 17921804 .
oster , g. , perelson , a. , katchalsky , a. , december 1971 .
network thermodynamics .
nature 234 , 393399 .
oster , g. f. , perelson , a. s. , katchalsky , a. , 1973 .
network thermodynamics : dynamic modelling of biophysical systems .
quarterly reviews of biophysics 6 ( 01 ) , 1134 .
paynter , h. m. , 1961 .
analysis and design of engineering systems .
mit press , cambridge , mass .
sengupta , b. , stemmler , m. , may 2014 .
power consumption during neuronal computation .
proceedings of the ieee 102 ( 5 ) , 738750 .
sengupta , b. , stemmler , m. , laughlin , s. b. , niven , j. e. , 07 2010 .
action potential energy efficiency varies among neuron types in vertebrates and invertebrates .
plos comput biol 6 ( 7 ) , e1000840 .
sengupta , b. , stemmler , m. b. , friston , k. j. , 07 2013 .
information and efficiency in the nervous system a synthesis .
plos comput biol 9 ( 7 ) , e1003157 .
shen , j. , yang , x. , dong , a. , petters , r. m. , peng , y .- w . , wong , f. , campochiaro , p. a. , 2005 .
oxidative damage is a potential cause of cone cell death in retinitis pigmentosa .
journal of cellular physiology 203 ( 3 ) , 457464 . smith , n. , crampin , e. , 2004 .
development of models of active ion transport for whole - cell modelling : cardiac sodium - potassium pump as a case study .
progress in biophysics and molecular biology 85 ( 2 - 3 ) , 387 405 .
sterratt , d. , graham , b. , gillies , a. , willshaw , d. , 2011 .
principles of computational modelling in neuroscience .
cambridge university press .
terkildsen , j. r. , niederer , s. , crampin , e. j. , hunter , p. , smith , n. p. , 2008 . using physiome standards to couple cellular functions for rat cardiac excitation - contraction .
experimental physiology 93 ( 7 ) , 919929 .
tran , k. , loiselle , d. s. , crampin , e. j. , 2015 .
regulation of cardiac cellular bioenergetics : mechanisms and consequences .
physiological reports 3 ( 7 ) , e12464 .
tran , k. , smith , n. p. , loiselle , d. s. , crampin , e. j. , 2009 .
a thermodynamic model of the cardiac sarcoplasmic / endoplasmic ca2 + ( serca ) pump .
biophysical journal 96 ( 5 ) , 2029 2042 .
ullah , g. , wei , y. , dahlem , m. a. , wechselberger , m. , schiff , s. j. , 08 2015 .
the role of cell volume in the dynamics of seizure , spreading depression , and anoxic depolarization .
plos comput biol 11 ( 8) , e1004414 .
van rysselberghe , p. , 1958 .
reaction rates and affinities . the journal of chemical physics 29 ( 3 ) , 640642 .
wei , y. , ullah , g. , schiff , s. j. , 2014 .
unification of neuronal spikes , seizures , and spreading depression .
the journal of neuroscience 34 ( 35 ) , 1173311743 .
wellstead , p. , 2012 . a new look at disease : parkinson s through the eyes of an engineer .
control systems principles , stockport , uk .
wellstead , p. , cloutier , m. , 2011 .
an energy systems approach to parkinson s disease .
wiley interdisciplinary reviews : systems biology and medicine 3 ( 1 ) , 16 .
wellstead , p. , cloutier , m. ( eds . ) , 2012 .
systems biology of parkinson s disease .
springer new york .
_ in vitro _
data was collected and analysed from retinal ganglion cells ( rgcs ) of wild - type ( wt ) ( n=8 ) and degenerative rd1 ( n=6 ) mice 4 - 4.5 month old .
experimental procedures were approved by the animal welfare committee at the university of melbourne and are in accordance with local and national guidelines for animal care .
animals were housed in temperature - regulated facilities on a 12h light / dark cycle in the animal house and had plentiful access to food and water .
neither wt nor rd1 mice were dark adapted for these experiments .
retinae from wt and rd1 mice were treated identically .
mice were anaesthetised with simultaneous ketamine ( 67 mg / kg ) and xylazine ( 13 mg / kg ) injections , the eyes were enucleated and then the mice were killed by cervical dislocation .
their eyes were bathed in carbogenated ( 95@xmath128 o2 and 5@xmath128 co2 ) ames medium ( sigma - aldrich , st .
louis , mo ) , hemisected at the ora serata , and the cornea and lens were removed .
the retina was continuously superfused with carbogenated ames medium at a rate of 4 - 8 ml / min .
all of the procedures were performed at room temperature and in normal room light .
the flat - mount retina was viewed through the microscope with the use of nomarski dic optics and also on a video monitor with additional 4x magnification using a ccd camera ( ikegami , icd-48e ) .
for whole cell recording , a small opening was first made with a sharp tip pipette ( resistance above 14 m@xmath129 ) through the inner limiting membrane and optic fibre layer that covered a selected retinal ganglion cell .
prior to recording , the pipette voltage in the bath was nullified .
the pipette series resistance was measured and compensated for using standard amplifier circuitry ( sec-05x ; npi electronic instruments ) .
pipette resistance was in the range of 7 - 14 m@xmath129 for all experiments .
membrane potentials were amplified ( as above with sec-05x , npi ) and digitised at 50 khz ( usb-6221 , national instruments ) , acquired and stored in digital form by custom software developed in matlab ( mathworks ) . to calculate the energy consumption in wt and rd1 rgcs
, hodgkin - huxley - type model parameters were fitted to the experimental data described above .
experimentally recorded maximum amplitude and width of spontaneous action potentials in two groups were averaged and used for model constraints .
a variety of voltage - gated ionic currents in rgcs have been identified experimentally : a calcium current ( @xmath130 ) , three types of potassium currents ( a - type ( @xmath131 ) , ca - activated ( @xmath132 ) , and delayed rectifier ( @xmath133 ) ) , t - type ca@xmath134 ( @xmath135 ) , hyperpolarization - activated ( @xmath136 ) and leakage ( @xmath137 ) currents .
the equation governing the membrane potential , @xmath2 , was obtained by summing all membrane currents using kirchoff s law , @xmath138 @xmath139 where @xmath140 a specific membrane capacitance .
the dynamics of each voltage - gated ionic current are governed by hodgkin - huxley - type gating variables , which are described by first - order kinetic equations as given in @xcite , but are omitted here for brevity . in this study , we sought to account for the differences in the energy consumption between wt and rd1 rgcs on the basis of differences in the magnitudes of the maximal conductance of sodium and potassium currents , @xmath141 and @xmath142 respectively . while all other parameters were kept fixed , @xmath141 and @xmath142 were systematically varied in the range [ @xmath143 in variable steps ( higher resolution for smaller values ) . for conservative calculation of the difference in the energy consumption ,
the smallest difference between @xmath141 and @xmath142 in wt and rd1 types that replicated experimental data is reported here .
model parameters used in simulations are given in table 5 ; @xmath144 , @xmath145 are maximum conductance and reversal potential of the current `` i '' . @xmath146 and @xmath147
depend on the intracellular ca@xmath134 concentration . at the moment ,
no experimental data is available on the properties of gating parameters for rgs in rd1 mice . due to this ,
the same gating parameters were used to model action potential in healthy and degenerative tissue .
a single - compartment hodgkin - huxley type neurons was simulated in neuron .
the standard euler integration method was used in simulations .
data was analysed in matlab ( mathworks ) .
l l + @xmath148 c & @xmath149 f/@xmath150 + @xmath151 mv & @xmath141 is varied in simulations + @xmath152 is variable , refer to @xcite & @xmath153 s/@xmath150 + @xmath154 mv & @xmath142 is varied in simulations + & @xmath155 s/@xmath150 + & @xmath147 is variable , refer to @xcite + @xmath156 mv & @xmath157 s/@xmath150 + @xmath158 mv & @xmath159 s/@xmath150 + @xmath160 mv & @xmath161 s/@xmath150 + the following simulation parameters reproduced the maximum amplitude and width of the experimentally recorded action potentials in wt and rd1 mice and gave the smallest difference in values between the two retina types .
parameters for wt : @xmath162 s/@xmath150 , @xmath163 s/@xmath150 .
parameters for rd1 : @xmath164 s/@xmath150 , @xmath163 s/@xmath150 . | energy - based bond graph modelling of biomolecular systems is extended to include chemoelectrical transduction thus enabling integrated thermodynamically - compliant modelling of chemoelectrical systems in general and excitable membranes in particular .
our general approach is illustrated by recreating a well - known model of an excitable membrane .
this model is used to investigate the energy consumed during a membrane action potential thus contributing to the current debate on the trade - off between the speed of an action potential event and energy consumption .
the influx of @xmath0 is often taken as a proxy for energy consumption ; in contrast , this paper presents an energy based model of action potentials .
as the energy based approach avoids the assumptions underlying the proxy approach it can be directly used to compute energy consumption in both healthy and diseased neurons .
these results are illustrated by comparing the energy consumption of healthy and degenerative retinal ganglion cells using both simulated and _ in vitro _ data .
biomolecular systems ; energy - based modelling ; excitable membranes ; retinal ganglion cells . |
investigation of nuclear systems with strangeness opens up a way to understand dense matter such as the neutron star core . since
strange quarks are negatively charged and cancel the positive proton charge , they are favored in charge neutral dense matter . as a result ,
many models predict hyperons would appear at around @xmath6 , and @xmath7 may share a similar or larger fraction to neutrons at very high densities . in describing highly dense matter , we clearly need information on @xmath8 interaction not only for @xmath9 but also for @xmath10 such as @xmath11 , but spectroscopic information on cascade ( @xmath5 ) hypernuclear systems are severely limited at present . while old emulsion data suggest a deep @xmath1-nucleus potential ( @xmath12 ) @xcite , a shallow potential ( @xmath13 ) is suggested from the twin hypernuclear event found in a nuclear emulsion @xcite .
this shallow potential is also supported by the distorted wave impulse approximation ( dwia ) analysis of @xmath1 production spectra in the bound state region @xcite . in order to extract as much information as possible from the data available at present , we need to investingate the @xmath1 hypernuclear production spectra by @xmath2 reaction on nuclear targets in both of the _ continuum _ as well as the bound state region .
since the dwia analysis in @xcite strongly rely on the absolute value of the inclusive @xmath1 production yield , it is necessary to verify the consistency with the production spectra in the quasi free ( qf ) region @xcite . for this purpose , the green s function method of dwia
would be a useful tool , where the continuum and bound state spectra can be described on the same footing . in this paper
, we investigate the @xmath1-nucleus potential through @xmath1 production spectra in the continuum and bound state region in the green s function method of dwia with the local optimal fermi averaging @xmath0-matrix ( lofat ) , in which the @xmath1-nucleus potential effects are included in both of the strength function and the transition amplitude .
based on the analyses of the observed continuum and bound region spectra , we make predictions to the future coming @xmath1 hypernuclear production experiment at j - parc .
we find that we should observe @xmath1 hypernuclear bound state peak structures in @xmath14 spectra on light nuclear targets such as @xmath3c and @xmath4al as far as the imaginary part of the @xmath1-nucleus optical potential is not large ( @xmath15 ) and the experimental resolution is good enough ( @xmath16 ) .
for @xmath7 and @xmath38 productions , it is recently pointed out that on - shell kinematics in the fermi averaging ( optimal fermi averaging , ofa ) procedure roughly decide the shape of the qf spectrum @xcite in the green s function method with factrized @xmath0-matrix , and similar procedure for @xmath0-matrix was proposed in ref .
@xcite . in the semi
classical distorted wave ( scdw ) analyses @xcite , the local fermi averaging of the elementray cross section has been included . here
we would like to incorporate both of the above two ideas ; we include the local optimal fermi averaging @xmath0-matrix ( lofat ) , @xmath39 , in the integrand of the response function eq .
( [ eq : response ] ) .
we define the lofat as , @xmath40 where @xmath41 denote the four total momenta in the elementary initial and final two - body states .
we adopt the fermi distribution function for the target nucleon momentum distribution @xmath42 and parameters are taken from @xcite . in obtaining lofat , we define the @xmath43-th hadron single particle energy containing the nuclear and hypernuclear potential effects as , @xmath44 this treatment enables us to include the potential effects naturally through the effective mass @xmath45 , as adopted in transport models in high - energy heavy - ion collisions @xcite .
consequently , the lofat has the dependence on the collision point @xmath46 through hadron potentials @xmath47 .
in the calculation , we have assumed the one body woods - saxon type hyperon - nucleus optical potential , @xmath48 , with @xmath49 , @xmath50 , @xmath51 , @xmath52 , where @xmath53 denotes @xmath1-core nucleus coulomb potential .
we assume the imaginary part of optical potential @xmath54 to be @xmath55 mev , which simulates the strength in the quark cluster model and the nijmegen potential model d @xcite estimations .
we have adopted the elementary @xmath0-matrix , which is re - parameterized to fit the cross section and angle dependence for @xmath56 3 gev / c .
figure [ figs : xi - qf ] shows the calculated results of @xmath1 qf production spectra with potential depth of 14 mev in comparison with experimental data @xcite .
calculated curves reproduce the experimental data systematically on heavy targets , cu , ag , and pb , in the high @xmath57 region , where the hypernuclear excitation is small . in the lower @xmath57 region , other contributions
have been known to be important @xcite , including heavy - meson production and its decay , @xmath58 ( @xmath59 ) @xcite and the two - step strangeness exchange and production processes , @xmath60 ( @xmath61 ) @xcite .
we underestimate the production spectra on lighter targets , @xmath3c and @xmath4al .
the underestimate of qf spectrum on @xmath3c target is a common feature in previous dwia calculations @xcite . in ref .
@xcite , it is discussed that this underestimate may be due to the center - of - mass effects : for electron scattering on a nucleus with mass number @xmath62 , the center - of - mass correction in the shell model have been taken care of by a multiplicative factor @xmath63 $ ] for the form factor , where @xmath64 is the momentum transfer . with @xmath65 , the factor @xmath66 ^ 2
$ ] amounts to 1.86 , 1.43 , 1.22 , 1.16 and 1.10 for @xmath3c , @xmath4al , @xmath67cu , @xmath68ag and @xmath69pb targets , respectively , at @xmath70 . in ref .
@xcite , we have adopted a different elementary @xmath0-matrix parameterization @xcite and larger isospin - averaged @xmath71 cross sections , then we can roughly explain the target mass dependence , as shown with the dashed lines in fig .
[ figs : xi - qf ] . since
our understanding is not complete and we have several ambiguities described above for the absolute yield in qf spectra , we introduce an adjustable multiplicative factor to fit the qf spectrum at low excitation energies ( high @xmath72 momentum region ) . in fig .
[ fig : qf - factor ] , we show the calculated @xmath1 production spectra with @xmath1 potential depth of @xmath73 and @xmath74 mev with multiplicative factors in comparison with data @xcite .
the potential depth dependence is small in low @xmath57 region , and attractive potential shifts the spectrum towards the high @xmath57 direction slightly .
when we adjust the multiplication factors as described above , calculated results reasonably well explain the qf spectrum .
this means that we can not determine the potential depth accurately from the qf spectrum shape . on the other hand ,
production spectrum is more sensitive to the potential depth at low excitation energies , then there is a possibility that we can determine the potential depth as discussed in refs .
@xcite . in fig .
[ figvdep ] , we show the results of the potential depth dependence of the calculated @xmath1 production spectrum on @xmath75c target at @xmath76 gev/@xmath77 multiplied by the factors described above in comparison with data @xcite .
_ evaluated the width of @xmath1 hypernuclear states ( @xmath78b+@xmath1 ) to possible double @xmath7 states based on the nijmegen model d potential as 1.2 mev and 0.5 mev for @xmath79^{j=1}$ ] and @xmath80^{j=0,2}$ ] , respectively @xcite . in the upper panel of fig .
[ figvdep ] , we show the ideal @xmath1 production spectra on @xmath75c target without the energy resolution folding . the imaginary part is assumed to be @xmath81 mev .
these calculated spectra show that the @xmath1 hypernuclear state widths are in good agreements with the estimates in ref . @xcite . in comparison with experimental data
, these spectra must be folded using a gauss function . in the lower panel of fig .
[ figvdep ] , we show the results with an experimental resolution of @xmath82 fwhm .
since the experimental resolution is not enough to distinguish the bound state peaks and the statistics is low , we should compare the integrated yield in the bound state region .
we find clear potential dependence in the bound state region , and with deep @xmath1 potential ( @xmath83 ) we may find a bump structure at around @xmath84 even with this low resolution . comparison with the data suggests that the potential depth around 14 mev is preferred in the present treatment , and if the statistics is high enough , it would be possible to determine the potential depth even with low resolution .
the depth of the @xmath1-nucleus potential has been already suggested to be around 14 mev from the analysis of the @xmath2 spectrum in the bound state region @xcite . in that analysis ,
the @xmath0-matrix element is evaluated under the frozen nucleon momentum approximation , where the kinematics is given with zero initial nucleon momentum . in the present analysis , while the kinematics in the elementary process and the way to fix the absolute value are different preferred potential depth is similar .
this may be due to the fact that the excitation energy dependence of lofat is weak and smooth since the covered @xmath72 momentum range is narrow in the bound state region .
now we find that @xmath1-nucleus potential with 14 mev depth well describes the spectra in the bound state and qf region for light nuclear targets , then it would be valuable to predict the peak structure which would be observed in the future coming j - parc day - one experiment .
sensitivity of calculated spectra for the @xmath1-nucleus potential is weak in the qf region because of the high momentum transfer @xmath85 500 mev / c , therefore it is difficult to extract precise potential information from the qf region , as shown in fig .
[ fig : qf - factor ] in fig .
[ bs ] , we show the calculated @xmath72 spectra in the bound state region of @xmath2 reactions on @xmath75c , @xmath86al , @xmath87ag and @xmath88pb targets with a potential depth of @xmath89 , which explains the qf spectra and low resolution spectra in the bound region .
we compare the results with @xmath90 ( solid lines ) and @xmath91 ( dotted lines ) .
we assume that the experimental resolution of @xmath92 would be achieved .
we find that bound state peaks are populated selectively due to high momentum transfer @xmath93 as in the @xmath7 production spectra by @xmath94 reactions ( @xmath95 ) , and these peaks can be identified in the high resolution experiment . in the green s function method , target nucleon
deep hole states are assumed to have large imaginary energies .
therefore , calculated results may be overestimating the spectra around the ground state due to the long lorentzian tail from the deep hole states having finite contributions in this energy region .
this problem will be discussed in the future .
we have investigated the @xmath1-nucleus potential through cascade ( @xmath5 ) hypernuclear production spectra by @xmath2 reaction in the green s function method @xcite of the distorted wave impulse approximation ( dwia ) with the local optimal fermi averaging @xmath0-matrix ( lofat ) @xcite .
the calculated spectra are in good agreement with the experimental data for heavy targets . with the multiplicative factor adjusted to fit the spectra on light targets , we find that the calculated spectra well reproduce the observed spectrum @xcite in shape and yield with @xmath1-nucleus potential depth around 14 mev .
this potential depth is consistent with those suggested in previous works @xcite . while the dependence on the potential depth is small in the continuum region , it is clealy distinguished in the bound region . therefore , it would be possible to extract the @xmath1-nucleus potential depth from the production yield in the bound state region when the statistics is high enough .
furthermore , the @xmath1 bound state peak structure can be found in the @xmath2 spectra on light target such as @xmath3c and @xmath4al , as far as the imaginary part is not very large ( @xmath96 ) and the experimental resolution is improved ( @xmath97 ) , as expected in the j - parc experiment .
we believe that our prediction would provide useful information in searching for the @xmath1 nuclear bound states at j - parc .
we would like to thank prof .
a. gal , prof . t. harada and prof .
m. kohno for valuable discussions .
this work is supported in part by the ministry of education , science , sports and culture , grant - in - aid for scientific research under the grant numbers , 15540243 , 1707005 , and 19540252 . c. b. dover and a. gal , annals phys .
* 146 * ( 1983 ) 309 .
s. aoki _ et al .
_ , phys .
b * 355 * ( 1995 ) , 45 .
t. fukuda _
et al . _ ,
phys . rev .
c * 58 * ( 1998 ) , 1306 .
p. khaustov _
et al . _ ,
c * 61 * ( 2000 ) , 054603 .
t. iijima _
et al . _ , nucl .
phys . a * 546 * ( 1992 ) , 588 .
e. h. auerbach , a. j. balitz , c. b. dover , a. gal , s. h. kahana , l. ludeking , and d. j. millener , ann .
( ny ) * 148 * ( 1983 ) , 381 .
o. morimatsu and k. yazaki , nucl .
phys . a * 483 * ( 1988 ) , 493 , _ ibid _ * 435 * ( 1985 ) , 727 . , m. t. lpez - arias , nucl .
phys . a * 582 * ( 1995 ) , 440 .
d. v. bugg _ et al .
_ , phys . rev .
* 168 * ( 1968 ) , 1466 .
s. tadokoro , h. kobayashi and y. akaishi , phys .
c * 51 * ( 1995 ) , 2656 .
b. w. allardyce _
et al . _ ,
a * 209 * ( 1973 ) , 1 .
t. harada and h. hirabayashi , nucl .
phys . a * 744 * ( 2004 ) , 323 ,
_ ibid _ * 767 * ( 2006 ) , 206 , _ ibid _ * 759 * ( 2006 ) , 143 .
y. alexander and p. j. moffa , phys .
c * 17 * ( 1978 ) , 676 .
et al . _ ,
* 112 * ( 2004 ) , 895 .
m. kohno et al . , phys .
c * 74 * ( 2006 ) , 064613 .
h. sorge , h. stoecker and w. greiner , annals phys .
* 192 * ( 1989 ) 266 ; a. b. larionov , w. cassing , c. greiner and u. mosel , phys .
c * 62 * ( 2000 ) 064611 ; m. isse et al .
, phys . rev .
c * 72 * ( 2005 ) 064908 [ arxiv : nucl - th/0502058 ] .
k. ikeda , t. fukuda , t. motoba , m. takahashi and y. yamamoto , prog .
* 91 * ( 1994 ) 747 . c. gobbi , c. b. dover , a. gal , phys .
c * 50 * ( 1994 ) , 1594 .
y. nara et al .
phys . a * 614 * ( 1997 ) , 433 . s. hashimoto , m. kohno , k. ogata and m. kawai , arxiv : nucl - th/0610126 . h. maekawa , k. tsubakihara , a. ohnishi , arxiv : nucl - th/0701066 . | we predict cascade hypernuclear production spectra expected in the forthcoming j - parc experiment . in the green s function method of the distorted wave impulse wave approximation with the local optimal fermi averaging @xmath0-matrix , we can describe the @xmath1 production spectra in the continuum and bound state region reasonably well .
predictions to the high resonlution spectra at j - parc suggest that we should observe @xmath1 bound state peak structure in @xmath2 spectra in light nuclear targets such as @xmath3c and @xmath4al .
@xmath5 hypernuclei , distorted wave impluse approximation , fermi averaging |
the detailed characterization of non - linear many - body quantum dynamics is a major goal of ultracold atomic and condensed matter physics . in recent years
there has been significant experimental progress in creating strongly interacting quantum systems within ultracold atomic ensembles ( see the review article @xcite and references therein , and recent work including @xcite ) .
the strong interaction between neutral alkali atoms and laser radiation allows for the prospect that such strongly interacting quantum systems can be probed and investigated at the resolution of single atoms .
this has been demonstrated very recently in two experiments @xcite ; it may be anticipated that the excellent results from these experiments will stimulate further experimentation using similar techniques .
the motivation to measure the spatial position of each atom in a strongly interacting quantum system is the desire to probe its structure and properties at an unprecedented level of detail . strongly interacting quantum systems tend to exhibit much more complex dynamics than weakly interacting or linear quantum systems
; this complexity leads to emergent behavior such as exotic phases of the quantum matter .
such behavior is often hard to simulate and fully understand on a theoretical level .
atom - by - atom measurements on these systems would allow much more detailed cross - checking of theoretical models against real data than is currently available , facilitating a much more complete understanding of these complicated dynamics .
the substantial flexibility available in choosing the system hamiltonian has led the idea that such systems may be used as ` quantum simulators ' to probe various many - body hamiltonians of interest in condensed matter physics @xcite . in order to take images of the atomic distribution at single - atom resolution , recent experiments
@xcite have used a deep far - detuned optical lattice to confine and mutually exclude the atoms as they are being measured . to detect the presence of an atom at a site with high confidence , the atoms must be localized to individual lattice sites throughout the duration of the measurement , despite substantial photon recoil heating .
this means that the atoms need to be cooled as they are fluorescing ; in current experiments polarization gradient cooling is used , with the cooling light collected to form the fluorescent signal .
if atoms were to hop between lattice sites , not only would the spatial resolution of the signal be degraded , but there is a good chance that atoms would be lost from the measurement region entirely by undergoing a light - assisted collision with another atom , or hopping to the edge of the lattice .
therefore , for efficient detection , hopping between lattice sites should be minimized while fluorescent scattering should be maximized .
the purpose of this article is to model the physical process of polarization gradient cooling of isolated atoms in a deep far - detuned optical lattice in order to understand and support future experiments performed using this technique .
in particular , the dependence of the atomic hopping rate between sites of the lattice is investigated as a function of experimental parameters .
the available literature concerned with the polarization gradient cooling of confined atoms is limited to the study of atomic dynamics in dissipative optical lattices ( for a comprehensive review , see @xcite ) , and does not directly address the situation at hand .
the polarization gradient cooling of ions has also been studied in separate work @xcite . in common with most theoretical treatments of laser cooling ,
this article uses a semi - classical monte carlo technique in order to model polarization gradient cooling in a computationally tractable manner .
the current situation differs from previous work in the context of dissipative optical lattices ( e.g. @xcite ) in the use of a separate far - detuned optical lattice to confine the atoms and a higher than usual atomic saturation parameter during the cooling process ; these both help to maximize the fluorescent scattering to hopping rate ratio .
this article also differs from previous articles by the use of a novel hybrid monte carlo - master equation ( hmcme ) technique to extract the quantity of interest the hopping rate between wells from the simulation .
this is necessary due to the difficulty in extracting the hopping rate from the monte carlo simulation alone .
the difficulty is due to the great disparity in time scales ; while intra - atomic dynamics take place on a timescale of @xmath1s , it is desired that atomic hopping between lattice sites takes place on a timescale greater than @xmath2s .
the direct monte carlo simulation of the hopping is therefore unfeasible .
the approach taken uses monte carlo simulations of the short - time dynamics to construct a master equation rate model for the simulation of the long - time behavior .
this is a novel approach for the simulation of ultracold atomic dynamics .
the present article studies the dynamics of atoms undergoing polarization gradient cooling in a deep far - detuned optical lattice .
while this is the basis for optical resolved - atom measurements of the spatial distribution of an atomic sample , other considerations are also important in optical resolved - atom experiments .
light - assisted collisions quickly eject atom pairs from multiply - occupied wells during the first small fraction of the measurement period ; in current experiments the parities of the site occupation numbers are measured rather than the true spatial distribution of the atoms .
another consideration is that it is desirable that the design of the optical apparatus allows the resolution of each individual lattice site within the imaging plane .
furthermore , if the atomic sample is three - dimensional , scattering from atoms in the imaging plane should be distinguishable from the scattering from out - of - focus atoms .
these considerations , while important , relate to the interpretation of the measurement data and to the design of the optical apparatus ; as such , they may be divorced from the basic physical process under examination in this article , and are left to be discussed elsewhere @xcite . looking beyond the two - dimensional measurements of the recent experiments @xcite , three - dimensional resolved - atom tomographic measurements of atomic distributions are possible with suitable apparatus @xcite .
tomographic measurements require multiple exposures to probe the complete atomic distribution ; this substantially increases the time needed for the measurement .
this places correspondingly stricter requirements on the atomic fluorescence and hopping rates , and it becomes much more important to optimize the scattering rate from each atom without recoil heating leading to atom loss during the measurement period .
the simulation developed in this article is used in a separate article by the author @xcite to inform the design of an experimental method which is capable of three - dimensional tomographic measurements of the position of each atom in strongly interacting quantum systems at half - wavelength spatial resolution . the article ( ref .
@xcite ) outlines a potential solution to the problems of resolvability and light - assisted collisions during fluorescent measurements of dense ultracold atomic systems .
furthermore , a solution to the out - of - plane scattering problem the background noise generated by fluorescent scattering from atoms outside the tomographic section in a three - dimensional sample is proposed , and is modeled using the simulation developed in the current article .
in fact the fluorescent imaging of atoms undergoing in - lattice polarization gradient cooling is a measurement technique which has the potential to be used in a wider class of experiments than just the resolved - atom experiments described above .
the ability to extract a very strong signal from each atom in a dilute sample may be useful in a variety of scenarios ; for example , to make a time - of - flight ( column - density ) measurement on a very low atom - number sample which is unobservable using absorption imaging due to its dilution .
unlike for resolved - atom measurements on dense samples , very high spatial resolution is likely not needed , so simpler apparatus may be used .
the structure of this article is as follows .
a conceptual outline of polarization gradient cooling for an atom in an optical lattice is given in section [ sdcoolconf ] , with emphasis on how the lattice affects the cooling process .
the system is described theoretically in section [ mcsimul ] , and the dynamical equations are rendered suitable for computation by the use of a semi - classical monte carlo wavefunction method . in section [ ext_mcsect ]
the hybrid monte carlo - master equation ( hmcme ) method is developed in order to predict the atomic hopping rate between wells . in section [ fricsect ] the frictional force felt by atoms undergoing polarization gradient cooling in a far - detuned optical lattice is calculated and compared to the case of untrapped atoms .
a calculation of the site hopping rate is presented for a specific set of parameters in section [ exampdyn ] .
the dependence of the site confinement time on the cooling configuration and system parameters is investigated in section [ lifehop ] , and conclusions are drawn .
polarization gradient cooling exploits the polarization gradients of light fields to cool atoms .
we briefly review two specific types of polarization gradient cooling which were analyzed in a seminal paper by dalibard and cohen - tannoudji @xcite : the one - dimensional orthogonal linear ( lin@xmath3lin ) and orthogonal sigma ( @xmath4 ) configurations for two equal intensity counter - propagating laser beams . for the lin@xmath3lin configuration , the positive component of the electric field is @xmath5 where @xmath6@xmath7@xmath8 .
the light potential consists of alternating standing waves of @xmath9 and @xmath10 light . for a ground - state atom with a spin of @xmath11=@xmath12 , the @xmath9 standing wave couples more strongly to the @xmath13=@xmath14 state , and vice versa ; the ground - state light shifts therefore oscillate alternately according to position ( see fig . [ nextcool1 ] ) .
furthermore , the @xmath9 ( @xmath10 ) light pumps atoms towards the @xmath13=@xmath14 ( @xmath13=@xmath15 ) state , which for negative laser frequency detuning , lies below the original state .
the motion of the atom is therefore described by a loss in translational energy as the atom climbs the state - dependent lattice potential , followed by optical pumping from a peak to a trough in the lattice potential accompanied by a much smaller recoil momentum exchange ( see fig .
[ nextcool1 ] ) .
this ` sisyphus ' process is dissipative . for the @xmath4 configuration ,
the positive component of the electric field is @xmath16 this polarization is always linear , but orientated at an angle which is proportional to @xmath17 ; it is convenient to think of a spatially varying atomic orientation axis parallel to the electric field , so the light is @xmath18 polarized everywhere . while the lights shifts are constant , they are not equal for ground state atoms with spin @xmath11@xmath19@xmath20 ; the clebsch - gordon coefficients are such that , with the laser tuned below resonance , the energy offset of the states increases monotonically with @xmath21 .
optical pumping preferentially transfers population from the high @xmath21 ( higher energy ) states to the low @xmath21 ( lower energy ) states . as the atom moves , the new basis is different to that of the original basis .
if the atom starts in a low @xmath21 state in the original basis , the atom has a higher proportion of its density matrix in the higher @xmath21 states in the new basis ; these lie higher in energy , and the atom feels a net force opposing its motion .
the optical pumping preferentially returns the atom into a low @xmath21 state at the new position , so the net force on an atomic trajectory is again dissipative
. + + for an atom confined to a region smaller than a wavelength the nature of the polarization gradient cooling processes are modified . in the cases under consideration
the atom is confined by an additional optical lattice potential , which shall be called the pinning lattice ; the pinning lattice has an optical frequency far away from atomic resonances , so the mixing of the ground and excited states induced by the pinning lattice potential is negligible compared to the mixing induced by the cooling beams , and consequently scattering from the lattice potential is negligible .
it is assumed that the pinning lattice potential is state - independent ( or nearly so ) for ground - state atoms , so the frequency detuning of the lattice is much greater than the width of the hyperfine structure of the excited state . in this section
we consider a pinning lattice formed by counter - propagating laser beams ; the lattice spatial period is @xmath22 , where @xmath23 is the wavelength of the lattice laser beams .
the counter - propagating lin@xmath3lin cooling light induces an additional state - dependent lattice potential with spatial period @xmath24 , where @xmath25 is the wavelength of the cooling light ( in this article the cooling and lattice laser wavelengths will be similar but not equal , @xmath26@xmath27@xmath28 ) .
the pinning lattice potential is characterized by its depth and phase with respect to the cooling light ( this phase will be a function of position if the lattice periods are not equal ) .
it is assumed that the pinning lattice is substantially deeper than the potential induced by the cooling light .
the pinning lattice may be in phase ( fig .
[ nextcool2 ] ) or out of phase ( fig . [ nextcool3 ] ) with the lattice potential produced by the cooling light
; these two situations have very different cooling characteristics . for the case of @xmath29=@xmath30 , there are equal intensities of @xmath9 and @xmath10 light at the minima of the combined potential , with equal and opposite intensity gradients , so the potential minima for @xmath13=@xmath31 are equal in energy , but offset a small distance from each other ( fig . [ nextcool3 ] ) . for laser light with a negative frequency detuning with respect to the atomic resonance
, atoms are optically pumped from the upper potential to the lower potential .
therefore for @xmath29=@xmath30 atoms are polarization gradient cooled efficiently at low energies as the optical pumping direction reverses near the potential minima ; sisyphus - like cycles can be performed even by deeply bound atoms . for the case of @xmath29=@xmath32 ,
the optical pumping changes sign half way up the pinning lattice potential ( fig . [ nextcool2 ] ) so sisyphus effects do not efficiently cool atoms with an energy below this point .
cooling can occur for atoms with energy greater than this point ; however calculations will show ( figure [ cooltv ] ) that this is less efficient than for the @xmath29=@xmath30 case .
the equilibrium temperature of atoms undergoing lin@xmath3lin polarization gradient cooling with @xmath29=@xmath32 will be much higher than for those with @xmath29=@xmath30 .
[ freqchandir ] another difference in the polarization gradient cooling of lattice - trapped atoms is that the additional pinning lattice potential rapidly changes the velocity and position of the atom .
if the intensity of the cooling light is such that the optical pumping time is greater than half of a lattice oscillation period , the atom can not undergo a full optical pumping cycle before the relative energies of the states are reversed by the motion of the atom ( see fig .
[ nextcool3 ] ) , so the cooling efficiency will be decreased if the cooling beams are not intense enough , and the steady - state temperature of the atoms is expected to increase at sufficiently low cooling intensity ( see figs .
[ varinten1d ] and [ varinten3d ] ) . in comparison , consider that without the lattice the atom encounters similar light fields at a rate of approximately @xmath33 , while the lattice - bound atom encounters similar light fields at a rate of approximately @xmath34 .
thus the intensity below which cooling becomes degraded is affected by the frequency of the lattice oscillations ; for deep enough lattices ( i.e. fast enough oscillations ) this intensity may be expected to be higher than is the case with no lattice present .
a further point is that the polarization gradient cooling process relies on inducing a differential dipole potential on the sublevels in order to generate the cooling force .
if the pinning lattice has a sublevel - dependent component then this modifies the cooling force , and can potentially disrupt the cooling process altogether .
the @xmath4 configuration in one dimension , unlike the lin@xmath3lin configuration , does not depend on the relative phase of the lattices as the light shift from the molasses is the same everywhere .
a change in phase of the molasses is identical to a rotation about the axis of the laser beams ; by a symmetry argument , it can be seen that the dynamics are unchanged in the direction parallel to the laser beams .
( there is a redistribution of the components of spontaneously emitted radiation in the orthogonal directions upon such a phase change , but the effect of this on the cooling dynamics is minor ) . as with the lin@xmath3lin configuration ,
the optical pumping should occur on a time scale shorter than half the period of oscillation for efficient cooling .
polarization gradient cooling can also take place if atoms are exposed to 3 sets of counter - propagating beams .
the atoms are exposed to both polarization and intensity gradients in 3d molasses light ; the polarization gradients cool the atoms by an admixture of the lin@xmath3lin and @xmath4 mechanisms @xcite .
tightly confined atoms interact with the cooling light within a volume smaller than a cubic wavelength , so the net intensity and polarization that an atom experiences depends on all five of the relative phases of the molasses beams . as a consequence the cooling efficiency along each of the three orthogonal directions at each pinning lattice site depends on the lattice site position together with these relative phases ; while some atoms will be cooled efficiently , others will be cooled only weakly .
furthermore atoms at different positions within the molasses light field will fluoresce at different rates ; this can pose a problem when correlating fluorescent signal to atom numbers .
this situation is unacceptable when the atoms are required , to a high probability , to be localized to each site for a long time .
this article discusses two ways to overcome this problem
. the first way would be to turn each set of counter - propagating one - dimensional @xmath4 beams on sequentially , cooling the atoms along each direction in turn .
the second way would be to introduce a small frequency difference between each pair of beams ; the phases of the cooling light at any point , and so the molasses character , will then change continuously .
this is the method used in recent experiments @xcite .
the two methods are analyzed in sections [ exampdyn ] and [ lifehop ] .
this section outlines the monte carlo simulation of the polarization gradient cooling of atoms confined in an additional optical potential .
the calculations presented in this article are performed for a single isolated atom , greatly simplifying the theoretical description of the system .
reabsorption of scattered radiation is possible amongst an extended atomic sample ; the influence of rescattering for a particular sample may be estimated by the ratio of the rescattered light to the incident light intensity at an atom .
however , the atom number and density will be low in the situations under consideration , so this ratio will typically be much less than unity
the object of the measurement is to determine the position of each atom in a sample , and this will not be possible if there are too many atoms , or they are too closely packed .
it is assumed that the additional confining optical potential has a frequency far away from the atomic resonance , so the mixing of the ground and excited states induced by the potential is negligible compared to the mixing induced by the cooling beams ; consequently fluorescent scatter from the optical potential is negligible . as a further consequence mixing between internal states induced by the optical potential
is ignored of course , this mixing must be present to generate the potential , but in the far - detuned regime this mixing is much smaller than the mixing due to the cooling beams , and is ignored . in this article
it is assumed that the additional optical potential is a three - dimensional optical lattice ( the pinning lattice ) , with three pairs of beams frequency detuned from each other by many megahertz , so the effective potential is the sum of the one - dimensional potentials .
the potential generated by the optical lattice is therefore @xmath35 where @xmath36 is a matrix dependent on the hyperfine structure of the transition ; in general it is state - dependent , and includes non - zero off - diagonal ground - ground or excited - excited elements dependent on the optical polarization .
the majority of alkali species have two ground hyperfine levels .
cooling takes place on a closed transition , i.e. @xmath37@xmath38@xmath39@xmath40@xmath41@xmath38@xmath42 , where @xmath43 is the nuclear spin ; however , the atoms also have a small probability of being excited to the @xmath41@xmath38@xmath44 state , which can decay to the lower hyperfine state @xmath37@xmath45@xmath44 .
the atoms will eventually be optically pumped to this dark state . as usual , this behavior is to be prevented by the presence of ` repumping ' light , i.e. light resonant with the @xmath37@xmath45@xmath39@xmath40@xmath41@xmath38@xmath44 transition .
it is assumed that this repumping light is of sufficient power that the total residence time in the @xmath37@xmath45@xmath44 state is only a few times the natural lifetime of the atom in the exited states ; consequently there is an extremely small population of atoms in the lower hyperfine state at any one time .
this means that the time spent in the lower hyperfine level is much less than the period of vibration in the lattice , so negligible heating of the atom will occur due to differences in the optical potentials of the two ground hyperfine states .
the repumping light is off - resonance by some gigahertz from the cooling transition @xmath37@xmath38@xmath39@xmath40@xmath41@xmath38@xmath42 , and as such will only have a very small effect on the dynamics of the atoms within those states , which is ignored in the subsequent analysis .
the population of the @xmath41@xmath38@xmath44 excited state is also ignored .
the hamiltonian describing the unitary evolution of the atomic state in the electric field of the cooling light in the rotating wave approximation is @xmath46 the state vector contains @xmath47@xmath38@xmath48 components describing the @xmath37@xmath38@xmath44 and @xmath41@xmath38@xmath42 states .
the matrix @xmath49 is the state - dependent potential , and @xmath50 the identity matrix .
the matrices @xmath51 ( @xmath52 ) are the dimensionless raising ( lowering ) matrices with values given by the clebsch - gordon coefficients .
these matrices are normalized so that @xmath53 in which @xmath54 is the excited state projection operator .
the relative normalization of the local electric field vector @xmath55 and the raman frequency parameter @xmath56 is arbitrary .
the frequency detuning from resonance of the cooling light is @xmath57 .
the evolution of the density matrix @xmath58 due to unitary processes obeys the von neumann equation @xmath59\ ; .\ ] ] the effects of the relaxation processes are determined from a trace over the ` environmental ' degrees of freedom of the corresponding unitary processes . to account for recoil from spontaneous emission processes ,
recoil momentum terms are added to the standard result @xcite in a sum over all transitions and polarizations @xmath60 here @xmath61 is a unit vector centered on the origin , and the parameter @xmath62 describes the angular distribution of the spontaneous emission of photons with polarization @xmath63 for a particular combination of raising and lowering operators @xmath64 and @xmath65 .
the magnitude of the laser wavevector is denoted by @xmath66 , and @xmath67 is the angular frequency width of the excited state .
the parameters @xmath62 are derived in appendix [ dir_sp ] . in many analyses of polarization gradient cooling
( e.g. @xcite ) the excited states are adiabatically eliminated .
this is appropriate for investigations that look to find the lowest temperature of atoms in optical molasses , as this usually occurs at low intensity ; at low intensities , the population of the excited states is small , and adiabatic elimination is a good approximation .
however , a high scattering rate is preferable for efficient signal extraction during the measurement process ; furthermore , as discussed in section [ freqchandir ] , the lowest temperatures of confined atoms tend to occur at higher intensity than for untrapped atoms .
therefore , the excited state populations are retained in the analysis in order to accurately simulate the situations of interest . in anticipation of taking the semi - classical approximation ,
the quantum dynamics of equations [ main_ham ] and [ sp_dyn_rec ] are expressed in terms of the wigner function @xmath68 it is shown in appendix [ spont_append ] that the complete dynamical equation in terms of the wigner function is @xmath69 where @xmath70 the hamiltonian @xmath71 is given by equation [ main_ham ] .
a full quantum treatment of the problem is computationally infeasible ; instead , the semi - classical approximation will be used . in the semi - classical approximation the external degrees of freedom of the atom are treated classically , while the internal degrees of freedom are treated quantum mechanically .
the semi - classical approximation is used in most theoretical treatments of polarization gradient cooling @xcite .
the approximation can be used in situations in which the coherence length of the atomic ensemble is much less than a wavelength ; it relies on the dephasing influence of spontaneous emission to in effect ` localize ' the atoms .
the accuracy of this approximation as used in the situation under consideration is discussed in section [ discusssimdyn ] . using the semi - classical approximation , the equations of motion
are expanded in the small parameter @xmath72 , with higher orders in the expansion discarded @xcite ; here @xmath73 is the momentum width of the wigner function .
for example , the expansion of the potential term on the right hand side of equation [ wig_comp ] is @xmath74 this quantity predominately contributes to the integral in the region @xmath75 .
the derivative of the potential @xmath76 has approximate magnitude @xmath77 , so the series expansion is seen to be in the small parameter @xmath78 ; it is terminated at appropriate order , which is chosen in this analysis to be the second order .
this approximation leads to the equation ( appendix [ ham_append ] ) @xmath79 + \frac{1}{2 } \sum_{i } \left \ { \frac{\partial { \bm w}}{\partial p_{i } } , \frac{\partial { \bm v}}{\partial r_{i } } \right \ } \\ & -&\frac{i \hbar}{8 } \sum_{ij } \left [ \frac{\partial^{2 } { \bm w}}{\partial p_{i}\partial p_{j } } , \frac{\partial^{2 } { \bm v}}{\partial r_{i}\partial r_{j } } \right ] + \gamma \sum_{\varepsilon}{\bm d^{\bm -}_{\bm \varepsilon } } { \bm w}{\bm d^{\bm + } _ { \bm \varepsilon } } \nonumber \\ \nonumber & -&\frac{\gamma}{2}\left ( { \bm p}{\bm w } + { \bm w}{\bm p } \right ) + \frac{\gamma \hbar^{2}k_{r}^{2}}{2}\sum_{\varepsilon,\varepsilon',i , j } \eta_{\varepsilon \varepsilon ' i j}{\bm d^{\bm -}_{\bm \varepsilon}}\frac{\partial^{2 } { \bm w}}{\partial p_{i}\partial p_{j}}{\bm d^{\bm + } _ { \bm \varepsilon ' } } \ ; .\end{aligned}\ ] ] in this equation , the curly brackets are the anti - commutator , and the tensor @xmath80 is given in appendix [ dir_sp ] .
the semiclassical evolution equation as it stands ( equation [ wig_app ] ) still requires substantial computational power to simulate . instead
, the calculation will be restricted to a single trajectory ; the distribution of atomic properties is then found by the sum over these trajectories .
this approach is stochastic i.e. monte carlo in nature .
the equations of motion for a specific trajectory in position and momentum space are found by substituting a semiclassical trial solution @xmath81 into equation [ wig_app ] .
this solution is valid in the limit in which both parameters @xmath72 and @xmath82 are small , as discussed in reference @xcite and in the previous section .
the parameter @xmath83 , where @xmath84 is the doppler temperature , is much smaller than one in the situations considered in this article .
integration over external co - ordinates gives the equation of evolution for the internal co - ordinates @xmath85+\gamma \left ( -\frac{1}{2}\left ( { \bm p}{\bm w } + { \bm w}{\bm p } \right ) + \sum_{\varepsilon } { \bm d^{\bm -}_{\bm \varepsilon}}{\bm \rho}{\bm d^{\bm + } _ { \bm \varepsilon } } \right ) \ ; .\ ] ] the equation of evolution of the external co - ordinates , e.g. @xmath86 , are found to be @xmath87 the equation containing @xmath88 describes the diffusion of the atom due to the atomic recoil .
the diffusion term is a fluctuating langevin force with zero mean , and is incorporated into the motion of a single trajectory in the it - langevin equation @xmath89 in which @xmath90 are independent , zero mean , gaussian - distributed stochastic increments with variance @xmath91 .
the quantities @xmath92 are the components of the @xmath93th eigenvector of @xmath88 , which is normalized according to its eigenvalue .
equations [ motion ] to [ ito_lang ] describe the motion of a single atom in the semi - classical approximation .
however , the internal dynamics as described in [ internal_app ] are still ensemble - averaged , and are not appropriate to describe a single trajectory .
the optical bloch equation ( equation [ internal_app ] ) must therefore be unraveled into the stochastic evolution of a single wave function .
the unraveling is chosen to be that of the quantum monte carlo wave function method ( qmcw ) @xcite .
the underlying principle of the approach is to evolve the wave function using a non - hermitian quasi - hamiltonian with the addition of randomly - occurring discrete quantum jump processes . in the qmcw approach
the following procedure is followed for each time step .
firstly , a random number 0@xmath94@xmath95@xmath941 chosen from a flat distribution and is compared to the quantity @xmath96 . if @xmath95@xmath94@xmath97 , latexmath:[\[\label{evol } if the @xmath95@xmath99@xmath97 , the atom undergoes a quantum jump ( photon emission ) ; photon polarization is chosen randomly according to the weights @xmath100 and the new wave function is given by @xmath101 the new wave function requires normalization in either case .
it can be shown that the ensemble of possible new wavefunctions satisfies the optical bloch equation ( equation [ internal_app ] ) .
the set of equations [ motion ] to [ proj_out ] specify the stochastic evolution of a single semiclassical trajectory .
an explicit third - order runge - kutta method is used to propagate the classical position and momentum of the atom .
the coefficients for the method are chosen so that the intermediate evaluation times are @xmath102 and @xmath103 , in which @xmath104 is the time step for the external dynamical evolution .
the internal components of the state vector are advanced at a constant time step @xmath102 , and are propagated by the cayley ( split ) form of the evolution operator .
the new wave function is found efficiently by gaussian elimination .
the method outlined in this section was tested against simple analytic models and by comparison against previously published work looking at polarization gradient cooling in one - dimensional dissipative optical lattices @xcite .
the results agree very well for intermediate saturation parameters . at high and low saturation parameters ,
the results differ somewhat ; the retention of the excited states means the method presented here is more accurate for high saturation parameters ( @xmath105@[email protected] ) , while the use of adiabatic elimination in the previous work enabled better statistics for lower saturation parameters ( @xmath105@[email protected] ) .
the monte carlo simulation was also tested against previous data for polarization gradient cooling of rubidium atoms in three - dimensional molasses @xcite ; for intermediate saturation parameters the predicted temperatures agreed to within 10% , which is around the error quoted for the previous work .
the problem under consideration differs from previous analyses of polarization gradient cooling in a number of respects .
the difference which poses the greatest computational challenge is the ratio of the time scales in the problem between the phenomena of interest , the jumping of atoms between wells , which we would like to take place over tens of seconds or greater ; and the smallest time scale relevant to the problem , the period of the beat frequency between the laser and atomic resonance , which will be in the range of 10@xmath107ns to 100@xmath107ns .
this means that , on average , one event of interest will occur every @xmath109 to @xmath110 time steps
. clearly a straightforward monte carlo simulation of these phenomena will be prohibitively slow .
adiabatic elimination is often used in simulations of polarization gradient cooling in order to increase the size of the smallest time step ( and decrease the size of the state vector ) .
however , this technique can only be used when the saturation parameter is much less than 1 , which is not the case for the situations under consideration in this article ( see sect . [ retentupper ] ) . in any case , the gain from this technique would be , at most , one or two orders of magnitude in the time scale ratio , i.e. interesting events would occur every @xmath111 to @xmath112 time steps ; such calculations would still be very computationally intensive . in order to make the simulation tractable
a method has been developed to extend the monte carlo simulation in order to study rare events . this technique bears similarity to previously developed techniques which are collectively known as the ` splitting ' methods of monte carlo simulation @xcite .
the basic idea is that a non - markovian system can look markovian when its dynamics are averaged over a sufficiently long time interval .
an atom undergoing fluorescent scattering is clearly non - markovian at timescales shorter than the scattering period ; the deterministic state - dependent dynamics of the bloch vector dominates , with stochastic dynamics ( environmental coupling ) playing only a minor role .
however , when viewed over sufficiently long time intervals , it is known ( from theoretical and experimental work ) that quantities such as temperature can be assigned to an atom undergoing polarization gradient cooling ; for such an assignment to make sense these quantities must be independent of the state of the atom at any one time , so implicitly it is assumed that the atomic dynamics , on long timescales , is described by stochastic markovian dynamics .
the hybrid monte carlo - master equation ( hmcme ) method presented in this section uses this idea to extend monte carlo simulation so that it is capable of predicting the frequency of rare events .
the method is approximate ; it uses the assumption that the system dynamics are markovian when viewed over appropriately long time intervals .
firstly , a quantity should be found that is representative of the aspect of the system which is to be investigated ; it should be a scalar time - dependent quantity which is a function of the state vector .
it is chosen with two properties in mind ; that it varies slowly and smoothly with time , and that the rare events of interest occur at values well separated from the usual values it takes .
for confined atoms undergoing polarization gradient cooling , an energy - like quantity is appropriate ; the atoms usually have energies well below that required to hop to a neighboring site .
in the context of the subsequent analysis , the approximation will be used that the system performs markovian dynamics a random walk along the energy axis .
the markovian approximation implies that the system is ergodic over long timescales .
it is desired to label the system according to a discrete energy parameter , so a set of ` points ' are defined at certain energies , @xmath113 . as
the time evolution is simulated using a monte carlo method , the system will encounter these points repeatedly .
( if the state vector of the system undergoes discontinuous jumps , the time and state vector of the system as it crosses a point may need to be extrapolated or approximated )
. the state of the system at any time will be classified by the last point encountered by the system .
each time the system encounters one of these points the state vector ( the complete description of the system ) is recorded ; these records will be called ` start vectors ' .
a representative set of start vectors are required for each point , i.e. a set which fairly samples the true population of start vectors for that point .
however , the run time of the monte carlo simulation should also be minimized .
as the system encounters points at higher energies only very rarely , it would be prohibitive to take data for these points by means of direct monte carlo simulation .
these two considerations need to be balanced . as a first stage
, the simulation is run from some arbitrary starting point until it settles down into a steady state condition . from this time on the state vector
is recorded every time the system encounters a point @xmath114 . a representative set of start vectors
will be built up for points at low energy .
this coverage is now extended to points lying at higher energy .
to do this a weight @xmath115 is assigned to each start vector ; this is an estimate of how representative that start vector is of the general population of start vectors at all points .
initially an equal weight is assigned to each of the extant start vectors ; the weights are normalized .
the total weight of the start vectors at a particular point @xmath116 is related , but not equal , to the probability of finding the steady state system in the environs of point @xmath117 .
the nature of the simulation is now changed so that new start vectors ( ` daughter ' start vectors ) are to be found from existing start vectors ( ` mother ' start vectors ) . to do this , a start vector is picked at random from all the possible start vectors according to its weight , and is evolved in time until a new point is reached . as it is assumed that the system is ergodic , the ` mother ' start vector at the old point can be replaced by the new ` daughter ' start vector at the new point ( by assigning the mother start vector s original weight to the daughter and setting the mother s weight to zero ) , and the overall set of start vectors is still representative of the steady state dynamics of the system . instead of the straight replacement of one mother
start vector by one daughter start vector , the simulation is performed multiple times for each mother start vector ; daughter vectors are found for the two neighboring points , together with the branching ratio .
the mother vector is replaced by two daughter vectors , randomly chosen from those calculated , with one at each neighboring point ; the weight of the mother vector is allocated to the daughters proportionately to the branching ratio .
it is now possible to focus the simulation on higher energies in the confining potential where the atom is found more rarely . once @xmath118 mother vectors have been propagated for a point @xmath114 , it is decided that enough data has been gathered to characterize the behavior of the system around this point , which is called ` full ' .
the start vectors at point @xmath114 are excluded from being propagated further using the monte carlo method ; the choice of mother vector to propagation is now determined from the weights of the start vectors at all points which are not full .
although no further daughter vectors are generated from full points , weight is still added to full points from neighboring partially - filled points ; if this goes uncompensated , it will lead to distortion of the distribution of weights between the points .
therefore as each mother vector is propagated and is replaced by daughter vectors , weight is reallocated amongst the full points , and from full points to partially - filled points , in order to preserve the relative allocation of weights between points .
details of this reallocation are given in appendix [ hmcme ] .
this method of extending the monte carlo simulation is not unique ; in practice the method outlined was found to give the best balance between reproducibility , accuracy and computation time for the problem in hand .
the method outlined in this section ( and summarized in appendix [ hmcme ] ) ensures that , as the simulation is run , it progresses from concentrating on the low energy region to the high energy region where the system is found only rarely . along
the way it has amassed data concerning the branching ratios between neighboring energy points ( the probability that a system , starting at one point , will end up at either neighbor ) together with the time taken to propagate from one point to the next .
it is now simple to calculate transfer rates between the points .
these transition rates are used in a master equation to calculate the relative populations of the atoms at various energies : @xmath119 where @xmath120@xmath7@xmath121 is the rate of transfer from @xmath122 to @xmath123 , @xmath124 the average time spent at @xmath122 , and @xmath125 the branching ratio .
it remains to find the actual quantity of interest
the hopping rate between sites .
the hopping rate of an atom at a particular point is directly extracted from the monte carlo data taken at that point ; the total hopping rate is calculated as the average hopping rate for all points weighted by the relative populations at those points .
the hop process is treated as a ` sink ' for the atomic population , i.e. once the atom has departed the well it is removed from the simulation .
a few subtleties were encountered when applying the algorithm to the polarization gradient cooling of confined atoms .
firstly , the potential energy of the system is ill - defined as the system has both quantum and classical properties .
the usual energy measure in such a scenario is the expectation value of the energy @xmath126 , where @xmath127 is the semi - classical position of the atom .
however , this energy measure does not satisfy the two conditions outlined in section [ measev ] , due to quantum jumps caused by the spontaneous emission of radiation , and by poor correlation between the value of the measure and the probability of site hopping ( take the example of an excited state atom stationary at a lattice minimum ) . instead
, another energy - type measure was used which has the required properties . using the kinetic energy as before ,
the potential energy was taken to be the lowest eigenvalue of the atomic system at the position of the atom .
the time evolution of this quantity is continuous ; furthermore , it correctly accounts for the combination of velocity and position needed to differentiate the main population of atoms from those which hop between sites .
the choice is further justified by noting that atoms undergoing polarization gradient cooling in a sublevel - independent lattice tend to be optically pumped to this lowest energy ground - state sublevel ( see sect .
[ orthlin ] ) .
the energy spacing of the points was in practice determined by a compromise between precision and computation time ; about 25 points were used spread over the energy range 0 to 1.3@xmath128 , where @xmath128 is the lattice depth . if more closely - spaced points were used , the computation proceeded faster , but at the expense of a larger spread in the prediction of the hopping rate between wells .
this was attributed to insufficient diffusion of the state vector as it is propagated from one point to the next leading to observable ` clumping ' of the simulated trajectories , and consequently undersampling of the phase space of start vectors at each point .
increasing the spacing between points , and so increasing the time the system spent at any particular point , decreased this trajectory clumping by increasing the influence of diffusive processes over shorter - time non - diffusive processes ; consequently this decreased the variation in the predicted jump rate between runs of the hmcme simulation at the expense of increased computation time .
around 150 start vectors were propagated for most points , rising to ten times this number near the lip of the lattice potential .
the accuracy of the algorithm presented in this section was tested by comparison with monte carlo simulations performed conventionally using the method described in section [ mcsimul ] ( which had been tested against previously published data see section [ compmeth ] ) .
there was excellent agreement between the hmcme method and the conventional monte carlo method throughout the region of overlap for all the data presented in this article ( e.g. figure [ expop ] ) .
to investigate the effect of the pinning lattice on the polarization gradient cooling of ultracold atoms it is instructive to compare the form of the frictional force with and without the lattice present .
conventionally the frictional force is calculated for an atom on a constant - velocity trajectory @xcite ; however this does not appropriately describe the motion of atoms moving in an additional lattice potential .
the average force and the average energy loss rate are instead calculated for an atom on a constant - energy trajectory .
as has been discussed in section [ implem ] , there is ambiguity in the definition of an energy measure in the semi - classical situation .
the measure used here for the potential energy ( as in the rest of this article ) is the position - dependent lowest - energy eigenvalue of the atom subject to both the pinning lattice and cooling light fields .
the average force and energy loss rates are calculated by integrating equations [ internal_app ] to [ motion2 ] using a density matrix method .
+ the force and energy loss rates for polarization gradient cooling without an additional lattice present are presented in figures [ coolfv ] and [ coolfe ] .
the dependence of the force on velocity for the @xmath4 configuration is very similar to previous analyses @xcite , while the lin@xmath3lin configuration looks somewhat different .
the differences for the lin@xmath3lin case are due to performing the calculation on a constant - energy rather than constant - velocity trajectory . for energies below 0.14@xmath129
the atom is confined to a single site of the optical lattice induced by the cooling light ( see fig . [ nextcool1 ] ) , so the atom is not cooled efficiently .
the atom experiences stronger cooling once it can move along the lattice . on the other hand ,
the induced dipole potential is spatially homogeneous in the @xmath4 configuration , so atoms moving on trajectories which have constant energy also have constant velocity ; results are found which are very similar to previous analyses .
if an energy measure is used which does not contain potential energy contributions from the cooling light , the form of the lin@xmath3lin velocity dependence alters to closely resemble that found in previous analyses ( e.g. @xcite ) .
it is worth noting that in practice atoms will follow neither constant - velocity nor constant - energy trajectories .
the cooling force and energy loss rate for atoms in a pinning lattice of depth 1.4@xmath130 are shown in figures [ cooltv ] and [ coolte ] .
the velocity plotted in figure [ cooltv ] is the velocity of an atom at the minima of the lattice potential .
the frictional energy loss profile for bound atoms is dependent on phase of the cooling beams at the pinning lattice site in the lin@xmath3lin configuration but not in the @xmath4 configuration .
as discussed in section [ orthlin ] , the cooling is very inefficient for atoms in a lattice with a @xmath29=@xmath32 phase relative to the cooling light , but is efficient in the @xmath29=@xmath30 case .
the character of the cooling is different for atoms which have an energy greater than the depth of the lattice .
the cooling of unbound atoms is similar to the cooling of atoms without the lattice present the lin@xmath3lin configuration becomes phase independent , while the @xmath4 configuration has greater cooling power .
on the other hand , atoms with energies around the lattice binding energy experience reduced cooling power .
this is due to time - averaging ; these atoms travel slowly when near the lip of the potential , so they are cooled more weakly at that time , and the average force falls .
these dips occur at slightly different positions due to the different depths of the overall potential in each case .
as discussed in section [ cool3dinit ] , three - dimensional molasses necessarily contains sub - wavelength scale intensity and polarization gradients . if an atom is confined so that it only samples the light field in a sub - wavelength region , the cooling efficiency is heavily dependent on the relative phases of the molasses light ( as demonstrated in section [ lat_pres ] ) .
experiments requiring reliably efficient three - dimensional cooling for tightly confined atoms need to find a way to overcome this problem .
two experimental scenarios are simulated in this section ; they differ in the method by which the phase problem is tackled
. one method ( which will be called the 1d alternating configuration ) uses a one - dimensional @xmath4 cooling beam configuration which is alternated between the three axes in turn .
this can be realized in an experiment by using square - wave pulses to modulate the input to three acousto - optical modulators . in the other method ( the 3d offset configuration ) , all six @xmath4 cooling beams are used at once , with a small frequency difference between each set of counter - propagating beams .
this sweeps the relative phases of the cooling light at each site .
the initial values of the relative phases of the cooling light at the pinning lattice site were chosen to be random for each run of the hmcme simulation .
the commonly used species @xmath131rb is chosen for the analysis ; similar results are expected to hold for other atomic species which are subject to polarization gradient cooling . the pinning lattice was chosen to be near the d@xmath132 atomic transition to allow efficient filtration of the intense lattice light from the weaker fluorescence light during the measurement .
the pinning lattice frequency detuning is chosen to be much greater than the hyperfine structure of the d@xmath132 line , and the lattice polarizations chosen to be linear , in order minimize the potential energy differences between the ground - state sublevels ( see section [ orthlin ] and @xcite ) . [ cols="<,<,^,^",options="header " , ] [ variousruns ] the results of the simulations predict that both the 1d alternating and the 3d offset cooling configurations produce excellent localization of a strongly - scattering atom at a pinning lattice site . however , the 3d offset configuration has a greater spread in the calculated temperatures , site hopping rates and photons scattering rates than the 1d alternating configuration ( compare the sample standard deviations in table [ paramsout ] ) .
this variability in the 3d offset configuration is primarily due to the role of the five relative phases of the cooling beams .
the [ 0,@xmath133 frequency detunings of the three sets of beams periodically vary the character of the cooling light at each pinning lattice site , but even so only samples a small portion of this 5-dimensional phase space , as only a single linearly independent phase parameter being varied in a cycle of frequency @xmath134 .
this means that changing the value of the four residual independent phases of the cooling beams does change the overall character of the light field at the lattice site , and this still affects the efficiency of the cooling at the lattice site , albeit to a lesser extent than if no frequency offset was used . to illustrate the residual variation in temperature and site lifetime , the worst confining set of phases for the 3d offset configuration
were further investigated from the data set investigated in section [ exampdyn ] ( which comprises 100 runs of the hmcme simulation at randomly chosen initial phases ) . the temperature and site lifetime for this set of phases
were found to be [email protected]@xmath107k and 10@xmath136([email protected])@xmath107s , showing substantially less confinement than the mean .
the photon scattering rate is also a function of these relative phases , as can be seen in the standard deviation of the scattering rate ( table [ paramsout ] ) , which is 8% of the mean value ; this variation is not good when performing accurate fluorescent measurements .
the variation in the cooling and photon scattering can be suppressed using a more complex set of frequency detunings between the cooling beams ; however , there is still likely to be some residual variation between lattice sites .
it is therefore concluded that for the situations that were simulated , while the 1d alternating and 3d offset configurations give similar temperatures and site hopping rates , the 1d alternating configuration exhibits less variability between different lattice sites and cooling laser phase arrangements , and hence is more suitable for use for fluorescent measurements .
it is seen that for a wide range of parameters the loss rate from a site of the pinning lattice due to radiative processes can be made to be @xmath137s@xmath138 or lower .
the loss rate of atoms from the whole lattice due to radiative processes will occur at a lower rate : to be lost from the lattice the atom would either need to encounter another atom and suffer an inelastic light - assisted collision , or travel to the edge of a lattice without being recaptured .
even so , the transfer of atoms between wells of the pinning lattice is unwanted as it acts to blur the spatial atomic distribution measurement .
regardless of the radiation scattering processes , the lifetime of atoms in the lattice is limited by collisions with room - temperature background gas particles , which inevitably lead to the loss of at least one atom from the lattice .
the lifetime due to background collisions varies by experiment and species , but typically is around @xmath2s . if the lifetime due to radiation scattering processes is greater than around @xmath139s , the loss rate due to such processes will be negligible compared to the loss rate due to background gas collisions , and it can be said that the polarization gradient cooling is optimal . in comparison ,
a typical image exposure time used in experiments is around 0.5@xmath107s to 1@xmath107s @xcite .
nevertheless , it is in general desirable to have the site lifetime as long as possible , which in practice means that it is limited by background gas collisions .
this is because the site lifetime has a direct bearing on the fidelity of the measurement : for a 1@xmath107s measurement , 1% of atoms are lost during the measurement if they have a 100@xmath107s lifetime , but around 10% are lost for a 10@xmath107s lifetime ; this is significant if hundreds of atoms are present in the image .
furthermore , and more perhaps importantly for future experiments , three - dimensional tomographic measurements have longer exposure times than two - dimensional measurements , as more images need to be taken ; consequently it is more critical that the site lifetime is as long as possible .
in fact the results show that the radiative site lifetime is several orders of magnitude above @xmath139s for a wide range of parameters
. this may be important for experiments as it adds to the robustness of the measurement technique .
consider that an additional source of heating is present , for example due to the reabsorption of scattered radiation ; the increased heating will tend to diminish the site lifetime exponentially .
if the parameters of the cooling light are chosen so there is a ` safety margin ' between the radiative site lifetime and the background gas collisional lifetime , the measurement technique can be made robust to the extra heating .
another very important consideration is that in an experiment it may be useful to use additional laser beams , on top of the ones discussed in this article , in order to aid measurements on the sample .
an example of this concept is discussed in a separate article @xcite , where it is proposed to use an additional laser beam to be able to distinguish between atoms undergoing fluorescent scatter at different depths in a three - dimensional sample ; this would be of clear benefit for tomographic measurements .
the additional probing beam adds substantial extra heating to the atoms , and it is beneficial to use near - optimal cooling parameters in order to diminish atom losses during the measurement .
the polarization gradient cooling of atoms trapped in a deep off - resonant optical lattice was simulated using a hybrid monte carlo - master equation technique .
the calculations indicate that there is a wide parameter range in which the lifetime at a lattice site of a strongly - scattering single atom undergoing polarization gradient cooling is primarily limited by collisions with background gas .
this is consistent with recent experiments @xcite .
the calculations suggest that for near - optimal parameters the radiative loss rate from a site is in fact orders of magnitudes less than the loss rate due to background gas collisions
. this could be important for experiments , as it suggests that the technique can be made robust in regards to additional heating mechanisms not included in the present analysis , for example heating from rescattered light or from additional probing beams .
it has been shown that by the use of an in - lattice polarization gradient cooling fluorescent measurement technique it is possible to extract large photon numbers of up to @xmath140 from each atom during fluorescent measurements of a dilute atomic sample .
this strongly supports the use of this technique when probing the spatial distribution of low atom - number samples of ultracold atoms .
the author would like thank christopher foot ( oxford , uk ) , members of the quantum processes and metrology group ( nist , gaithersburg , usa ) , and the nist internal referees for their comments , and to acknowledge funding from the epsrc ( uk ) , christ church college , oxford ( uk ) , the lindemann trust ( uk ) and the atomic physics division of nist , gaithersburg ( usa ) .
the intensity of the spontaneously scattered radiation is ( @xcite p.173 ) @xmath141 the electric field of the scattered light in the far field is determined from the source - field expression ( @xcite p.328 ) @xmath142 in which the unit vector @xmath143 is the polarization vector of the radiation mode labeled by @xmath144 . dividing the intensity into two perpendicular polarizations , conveniently @xmath145 and @xmath146 in spherical polar coordinates ,
the following relationships are obtained : @xmath147 in which mode @xmath144 has been identified by the unit wave vector @xmath61 and the polarization .
the expectation value runs over the internal degrees of freedom of the atom . by breaking down the dipole operator into the component basis ,
the intensity is expressed in the form @xmath148 the coefficients @xmath149 are listed in table [ f_params ] with the normalization chosen so that the total scattered power is @xmath150 .
the tensor @xmath80 of equation [ wig_app ] is related to the functions @xmath149 by @xmath151 with @xmath152 being the unit vector in the @xmath117 direction .
the tensor @xmath80 is given in table [ eta ] .
b|aaa ' '' '' f^_ & ' '' '' =-1 & ' '' '' =0 & ' '' '' =1 + ' '' '' = -1 & ' '' '' ^2 & ' '' '' - 2 & ' '' '' - ^2 + ' '' '' = 0 & ' '' '' - 2 & ' '' '' ^2 & ' '' '' 2 + ' '' '' = 1 & ' '' '' - ^2 & ' '' '' 2 & ' '' '' ^2 + b|aaa ' '' '' f^_ & ' '' '' =-1 & ' '' '' =0 & ' '' '' =1 + ' '' '' = -1 & ' '' '' & ' '' '' 0 & ' '' '' + ' '' '' = 0 & ' '' '' 0 & ' '' '' 0 & ' '' '' 0 + ' '' '' = 1 & ' '' '' & ' '' '' 0 & ' '' '' + [ f_params ] b|aaa ' '' '' _ i j & ' '' '' =-1@xmath153 & ' '' '' =0@xmath153 & ' '' '' =1@xmath153 + ' '' '' = -1 & ' '' '' & ' '' '' & ' '' '' + = 0 & ' '' '' & ' '' '' & ' '' '' + = 1 & ' '' '' & ' '' '' & ' '' '' +
the transformation of the kinetic energy operator term of the unitary dynamics ( equation [ main_ham ] ) is covered in standard treatments of the derivation of the wigner - moyal equation @xcite .
the transformation of the potential energy operator term of equation [ main_ham ] proceeds straightforwardly using the position space representation of the wigner transformation ( equation [ wiga ] ) . for the non - unitary dynamics of equation [ sp_dyn_rec ] , the transformation of the terms containing the constant matrix @xmath54 is trivial .
the remaining term of the equation is @xmath154 for this term it is easier to use the wigner transformation as expressed in the momentum representation , which is found to be @xmath155 the application of this transform onto the expression [ sp_dyn_rec2 ] involves expressions such as @xmath156 which are evaluated by identification of the momentum translation operator .
the matrices @xmath157 are constant in spatial coordinates , so the wigner transformed term is @xmath158
the semiclassical approximation is taken by the termination of the expansion ( equation [ wig_exp ] ) of the right hand side of equation [ wig_comp ] in terms of small changes in momenta .
terms containing polynomials in @xmath159 are expressed in terms of the derivative of the wigner function with respect to momentum , for example : @xmath160 the recoil term of equation [ wig_comp ] is directly expanded in terms of small momenta : @xmath161 integrals are performed over the unit sphere , and a sum taken over polarizations .
terms containing odd powers of kappa vanish , leaving a residual second order term @xmath162 the tensor @xmath80 defined in equation [ closc7 ] .
the conversion of equation [ wig_app ] to langevin form is achieved by substitution of the trial solution of equation [ delta_app ] . to find the equation of motion for the internal coordinates ( equation [ internal_app ] )
the substitution is performed directly and the integral taken over the external co - ordinates . to find the equations of motion for the external co - ordinates ( equations [ motion ] to [ motion_end ] ) the trial solution
is substituted and the appropriate expectation value taken ( by integration over both internal and external coordinates ) .
a key relation concerns the derivatives of the delta distribution @xmath163 using this relation the equations of motion are found for @xmath164 etc .
the non - zero expectation values for the second moments of momentum are interpreted as a langevin force .
a measure is chosen which is representative of the dynamics of interest ( see section [ measev ] ) , and a series of points are chosen on this measure . the value of the measure varies in time as the system evolves .
the system is classified at any time according to the last point encountered by the measure . when the system s classification changes ( when the measure encounters a new point ) , the state vector of the system is recorded ; this is called a ` start vector ' for the new point . 1 . 1 . one or more
initial standard monte carlo simulations are carried out .
after the system settles into a steady state , the state vector is recorded every time the system encounters a new point ; these form the initial population of start vectors for the rest of the hmcme simulation .
this continues until enough start vectors have been found to characterize a few of the points on the measure ( those where the system is most commonly found ) .
an equal weight is assigned to each of the initial start vectors .
2 . 1 . a single start vector ( the mother )
is chosen randomly from the set of all start vectors at partially filled points with a probability which is proportional to the start vector s weight @xmath165 .
2 . the mother start vector is propagated multiple times using the monte carlo method until the next point is reached .
the resulting state vectors ( the daughters ) are categorized by end point ; a single randomly chosen daughter vector is retained for each end point .
the branching probabilities @xmath166 for the mother start vector are calculated .
3 . weights @xmath167@xmath7@xmath168 are assigned to the two daughter vectors .
if one of the daughter vectors is at a full point , the daughter s weight is added to the combined weight at that point .
4 . for each full point
( with weight @xmath169 ) a new weight is assigned according to @xmath170 with @xmath171 the total weight at all partially filled points . the weight from this point
is distributed to the two neighboring points .
if a neighboring point ( with weight @xmath172 ) is also full , it is assigned the new weight @xmath173 where @xmath174 are the calculated mean branching ratios from @xmath97 to @xmath93 . if the neighboring point @xmath93 is not full the start vectors at that point which originate from point @xmath97 are assigned the weights @xmath175 where the summation @xmath176 runs over all the daughter vectors of the point @xmath97 at point @xmath93 .
the reallocation of weights described in this step is carried out so weight is reallocated from each full point .
a weight @xmath177@xmath7@xmath178 is assigned to the mother vector .
if the number of start vectors propagated reaches the maximum allowed @xmath118 at a point @xmath117 , the point is called ` full ' ; the weights @xmath179 of all start vectors at this point are added to give @xmath180 , and the mean branching ratios @xmath166 are found for all relevant channels . 7 .
part 2 is repeated until data has been taken for all points of interest .
the results of the part 2 are processed to form a master equation as discussed in section [ marknonmark ] .
the reasoning behind the reallocation of weights from the full points ( step 2(d ) ) is to prevent a false buildup of weights at the full points ; in the absence of this reallocation , weight is transferred from partially filled to full points , but not the other way around . to compensate for not propagating vectors from these full points ,
the weights are reallocated ( equations [ d1 ] to [ d3 ] ) in the same proportion that would occur if start vectors were propagated from the full points .
these equations are derived by noting that , when each start vector is propagated , a fraction @xmath181 of the weight of @xmath182 ( the total weight at partially filled points , i.e. the total weight from which the start vector is chosen ) is being reallocated ; therefore , to compensate for not propagating start vectors from the full points , an equivalent fraction of the weight at full points @xmath183 , for all full points @xmath97 , should be reallocated according to the known branching ratios . | the atom - by - atom characterization of quantum gases requires the development of novel measurement techniques .
one particularly promising new technique demonstrated in recent experiments uses strong fluorescent laser scattering from neutral atoms confined in a short - period optical lattice to measure the position of individual atoms in the sample .
a crucial condition for the measurements is that atomic hopping between lattice sites must be strongly suppressed despite substantial photon recoil heating
. this article models three - dimensional polarization gradient cooling of atoms trapped within a far - detuned optical lattice .
the atomic dynamics are simulated using a hybrid monte carlo and master equation analysis in order to predict the frequency of processes which give rise to degradation or loss of the fluorescent signal during measurements .
it is shown , consistent with the experimental results , that there exists a wide parameter range in which the lifetime of strongly - fluorescing isolated lattice - trapped atoms is limited by background gas collisions rather than radiative processes . in these cases
the total number of scattered photons can be as large as @xmath0 per atom .
the performance of the technique is related to relevant experimental parameters . |
the shape of a set of points , the shape of a signal , the shape of a surface , or the shapes in an image can be defined as follows : the remainder after we have filtered out the position and the orientation of the object @xcite . statistics on shapes appear in many fields .
paleontologists combine shape analysis of monkey skulls with ecological and biogeographic data to understand how the _ skull shapes _ have changed in space and time during evolution @xcite .
molecular biologists study how _ shapes of proteins _ are related to their function .
statistics on misfolding of proteins is used to understand diseases , like parkinson s disease @xcite .
orthopaedic surgeons analyze _
bones shapes _ for surgical pre - planning @xcite . in signal processing ,
the _ shape of neural spike trains _ correlates with arm movement @xcite . in computer vision , classifying _ shapes of handwritten digits _ enables automatic reading of texts @xcite . in medical imaging and more precisely in neuroimaging , studying _ brain shapes _ as they appear in the mris facilitates discoveries on diseases , like alzheimer @xcite .
what do these applications have in common ?
position and orientation of the skulls , proteins , bones , neural spike trains , handwritten digits or brains do not matter for the study s goal : only _ shapes _ matter .
mathematically , the study analyses the statistical distributions of _ the equivalence classes of the data _ under translations and rotations .
they project the data in a quotient space , called the _ shape space_. the simplest - and most widely used - method for summarizing shapes is the computation of the mean shape .
almost all neuroimaging studies start with the computation of the mean brain shape @xcite for example .
one refers to the mean shape with different terms depending on the field : mean configuration , mean pattern , template , atlas , etc .
the mean shape is an average of _ equivalence classes of the data _ : one computes the mean after projection of the data in the shape space .
one may wonder if the projection biases the statistical procedure .
this is a legitimate question as any bias introduced with this step would make the conclusions of the study less accurate .
if the mean brain shape is biased , then neuroimaging s inferences on brain diseases will be too .
this paper shows that a bias is indeed introduced for the mean shape estimation under certain conditions .
we review works on the shape space s geometry as a quotient space , and existing results on the mean shape s bias .
* shapes of landmarks : kendall analyses * the theory for shapes _ of landmarks _ is introduced by kendall in the 1980 s @xcite .
he considers shapes of @xmath1 labeled landmarks in @xmath2 .
the size - and - shape space , written @xmath3 , takes also into account the overall size of the landmarks set . the shape space , written @xmath4 , quotients by the size as well .
both @xmath3 and @xmath4 have a riemannian geometry , whose metrics are given in @xcite .
these studies model the probability distribution of the data directly in the shape space @xmath4 .
they do not consider that the data are observed in the space of landmarks @xmath5 and projected in the shape space @xmath4 .
the question of the bias is not raised .
* shapes of landmarks : procrustean analyses * procrustean analysis is related to kendall shape spaces but it also considers shapes of landmarks @xcite .
kendall analyses project the data in the shape space by explicitly computing their coordinates in @xmath4 .
in contrast , procrustean analyses keep the coordinates in @xmath5 : they project the data in the shape space by `` aligning '' or `` registering '' them .
orthogonal procrustes analysis `` aligns '' the sets of landmarks by rotating each set to minimize the euclidean distance to the other sets . procrustean analysis considers the fact that the data are observed in the space @xmath5 but does not consider the geometry of the shape space . the bias on the mean shape is shown in @xcite with a reducto ad absurdum proof .
but there is no geometric intuition given about how to control or correct the phenomenon .
* shapes of curves * the curve data are projected in their shape space by an alignment step @xcite , in the spirit of a procrustean analysis .
the bias of the mean shape is discussed in the literature .
unbiasedness was shown for shapes of signals in @xcite but under the simplifying assumption of no measurement error on the data .
some authors provide examples of bias when there is measurement error @xcite .
their experiments show that the mean signal shape may converge to pure noise when the measurement error on simulated signals increases .
the bias is proven in @xcite for curves estimated from a finite number of points in the presence of error .
but again , no geometric intuition nor correction strategy is given .
we are missing a global geometric understanding of the bias . which variables control its magnitude ?
is it restricted to the mean shape or does it appear for other statistical analyses ?
how important is it in practice : do we even need to correct it ?
if so , how can we correct it ?
our paper is addressing these questions .
we use a geometric framework that unifies the cases of landmarks , curves , images etc .
[ [ contributions ] ] contributions + + + + + + + + + + + + + we make three contributions .
first , we show that statistics on shapes are biased when the data are measured with error .
we explicitly compute the bias in the case of the mean shape .
second , we offer an interpretation of the bias through the geometry of the shape space . in applications ,
this aids in deciding when the bias can be neglected in contrast with situations when it must be corrected .
third , we leverage our understanding to suggest several correction approaches .
[ [ outline ] ] outline + + + + + + + the paper has four sections .
section 1 introduces the geometric framework of shape spaces .
section 2 presents our first two contributions : the proof and geometric interpretation of the bias .
section 3 describes our third contribution : the procedures to correct the bias .
section 4 validates and illustrates our results on synthetic and real data .
we introduce two simple examples of shape spaces . we will refer to them constantly to provide intuition . first , we consider two landmarks in the plane @xmath6 ( figure [ fig : simple ] ( a ) ) .
the landmarks are parameterized each with 2 coordinates .
for simplicity we consider that one landmark is fixed at the origin on @xmath6 .
thus the system is now parameterized by the 2 coordinates of the second landmark only , e.g. in polar coordinates @xmath7 .
we are interested in the shape of the 2 landmarks , i.e. in their distance which is simply @xmath8 .
second , we consider two landmarks on the sphere @xmath9 ( figure [ fig : simple ] ( b ) ) .
one of the landmark is fixed at the origin of @xmath9 .
the system is now parameterized by the 2 coordinates of the second landmark only , i.e. @xmath10 . the shape of the two landmarks is the angle between them and is simply @xmath11 .
the data are objects @xmath12 that are either sets of landmarks , curves , images , etc .
we consider that each object @xmath13 is a point in a riemannian manifold @xmath14 .
we restrict in this paper to finite dimensional manifolds in order to avoid complications .
we have @xmath15 in the plane example : a flat manifold of dimension 2 .
we have @xmath16 in the sphere example : a manifold of constant positive curvature and of dimension 2 . by definition ,
the objects shapes are their equivalence classes @xmath17\}_{i=1}^n$ ] under the action of some finite dimensional lie group @xmath18 : @xmath18 is a group of continuous transformations that models what does not change the shape . the action of @xmath18 on @xmath14 will be written with `` @xmath19 '' . in our examples ,
the rotations are the transformations that leave the shape of the systems invariant .
let us take @xmath20 a rotation .
the action of @xmath20 on the landmark @xmath21 is illustrated by a blue arrow in figures [ fig : leadingcases1 ] ( a ) for the plane and ( d ) for the sphere .
we observe that the action does not change the shape of the systems : the distance between the two landmarks is preserved in ( a ) , the angle between the two landmarks is preserved in ( d ) .
the equivalence class of @xmath13 is also called its orbit and written @xmath22 .
the equivalence class / orbit of @xmath21 is illustrated with the blue dotted circle in figure [ fig : leadingcases1 ] ( a ) for the plane example and in figure [ fig : leadingcases1 ] ( b ) for the sphere example .
the orbit of @xmath21 in @xmath14 is the submanifold of all objects in @xmath14 that have the same shape as @xmath21 .
the curvature of the orbit as a submanifold of @xmath14 is the key point of the results in section [ sec : quant ] .
the _ shape space _ is by definition the space of orbits .
this is a quotient space denoted @xmath23 .
one orbit in @xmath14 , i.e. one circle in figure [ fig : leadingcases1 ] ( b ) or ( e ) , corresponds to a point in @xmath24 .
the shape space is @xmath25 in the plane example .
this is the space of all possible distances between the two landmarks , see figure [ fig : leadingcases1 ] ( c ) .
the shape space is @xmath26 $ ] in the sphere example .
this is the space of all possible angles between the two landmarks , see figure [ fig : leadingcases1 ] ( f ) .
we consider that the action of @xmath18 on @xmath14 is _ isometric with respect to the riemannian metric of @xmath14_. this implies that the distance @xmath27 between two objects in @xmath14 does not change if we transform both objects in the same manner . in the plane example , rotating the landmark @xmath28 and another landmark @xmath29 with the same angle does not change the distance between them .
the distance in @xmath14 induces a quasi - distance @xmath30 in @xmath24 : @xmath31 .
the distance between the shapes of @xmath28 and @xmath29 is computed by first registering / aligning @xmath28 onto @xmath29 by the mean of @xmath20 , and then using the distance in the ambient space @xmath14 . in the plane example
, the distance between two shapes is the difference in distances between the landmarks .
one can compute it by first aligning the landmarks , say on the first axis of @xmath6 .
then , one uses the distance in @xmath6 . both object space @xmath14 and shape space @xmath24 are stratified because of the notion of isotropy group .
the _ isotropy group of @xmath13 _ is the subgroup of transformations of @xmath18 that leave @xmath13 invariant .
for the plane example , every @xmath32 has isotropy group the identity and @xmath33 has isotropy group the whole group of 2d rotations .
objects on the same orbit , i.e. objects that have the same shape , have conjugate isotropy groups .
the _ orbit type _ of an orbit is the corresponding conjugate class . _
principal shapes _ are shapes with smallest isotropy group conjugation class . in the plane example
, @xmath34 is the set of objects with principal shapes .
it corresponds to @xmath35 in the shape space and is colored in blue on figure [ fig : leadingcases1 ] ( c ) .
_ singular shapes _ are shapes with larger isotropy group conjugation class . in the plane example
, @xmath33 is the only object with singular shape .
it corresponds to @xmath36 in @xmath37 and is colored in red in figure [ fig : leadingcases1 ] ( c ) .
the ( connected components of ) the _ orbit types _ form a stratification of @xmath14 , called the _ orbit - type stratification of @xmath14_. the principal type is predominant in the following sense : the set of principal strata , which we denote @xmath38 , is open and dense in @xmath14 .
this means that there are objects with non - degenerated shapes almost everywhere .
the stratification of @xmath14 into orbit types strata gives a stratification of the shape space @xmath39 .
also , non - degenerated shapes are dense in the shape space .
we have focused on an intuitive introduction of the concepts .
we refer to @xcite for mathematical details . from now on ,
the mathematical setting is the following : we assume a proper , effective and isometric action of a finite dimensional lie group @xmath18 on a finite dimensional riemannian manifold @xmath14 .
we recall that the data are the @xmath12 that are sets of landmarks , curves , images , etc .
we interpret the data @xmath13 s as random realizations of the generative model : @xmath40 where the observed object @xmath41 is a shape @xmath42 with a given position or parameterization @xmath43 and observed with noise @xmath44 . here
exp(p , u ) denotes the riemannian exponential of @xmath45 at point @xmath46 .
the @xmath47 are themselves i.i.d .
realizations of random variables . drawing them
lead to the following three step interpretation of the generative model [ eq : genmodel ] .
[ [ step-1-generate - the - shape - y_i - in - mg ] ] step 1 : generate the shape @xmath48 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we assume that there is an probability density of shapes in @xmath23 , with respect to the measure on @xmath24 induced by the riemannian measure of @xmath14 .
the @xmath49 s are i.i.d .
samples drawn from this distribution .
for example , it can be a gaussian as illustrated in figure [ fig : genmodel1 ] on the shape spaces for the plane and sphere examples .
this is the variability that is meaningful for the statistical study , whether we are analyzing shapes of skulls , proteins , bones , neural spike trains , handwritten digits or brains .
we assume in this paper that the distribution is simply a dirac at @xmath50 which we call the _ template shape_. this is the most common assumption in these generative models @xcite .
[ [ step-2-generate - its - positionparameterization - g_i - in - g ] ] step 2 : generate its position / parameterization @xmath43 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we can not observe shapes in @xmath23 .
we rather observe objects in @xmath14 , that are shapes posed or parameterized in a certain way .
we assume that there is a probability distribution on the positions or parameterizations of @xmath18 , or equivalently a probability distribution on principal orbits with respect to their intrinsic measure .
we assume that the distribution does not depend on the shape @xmath49 that has been drawn .
the @xmath51 s are i.i.d . from this distribution .
for example , it can be a gaussian as illustrated in figure [ fig : genmodel2 ] on the shape spaces for the plane and sphere examples .
[ [ step-3-generate - the - noise - epsilon_i - in - t_g_i - cdot - y_im ] ] step 3 : generate the noise @xmath44 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the observed @xmath13 s are results of noisy measurements .
we assume that there is a probability distribution function on @xmath52 representing the noise .
we further assume that this is a gaussian centered at @xmath53 , the origin of the tangent space @xmath52 , and with standard deviation @xmath0 , see figures [ fig : genmodel3 ]
. the parameter @xmath0 will be extremely important in the developments of section , as we will compute taylor expansions around @xmath54 .
other generative models may be considered in the literature .
we find in @xcite the model : @xmath55 and in @xcite the model : @xmath56 .
our goal is to unveil the variability _ of shapes in @xmath23 _ while we in fact observe the noisy _ objects @xmath13 s in @xmath14_. first , we focus on the case where the variability in the shape space is assumed to be a dirac at @xmath50 ( step 1 of generative model ) .
our goal is thus to estimate the template shape @xmath50 .
one may consider the maximum likelihood estimate of @xmath50 : @xmath57 we have hidden variables , the @xmath20 s .
the expectation - maximization ( em ) algorithm would be the natural implementation for computing the ml estimator . but the em algorithm is computationally expensive , above all for tridimensional images .
thus , one usually relies on another procedure that is an approximation of the em .
@xcite .
[ [ estimating - the - template - shape - with - the - frchet - mean - in - the - shape - space ] ] estimating the template shape with the frchet mean in the shape space + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + one initializes the estimate with @xmath58 .
then , one iterates the following two steps until convergence : @xmath59 ( 1 ) is an estimation of the hidden observations @xmath51 and an approximation of the e - step of the em algorithm .
( 2 ) is the m - step of the em algorithm : the maximization of the surrogate in the m - step amounts to the maximization of the variance of the projected data .
this is exactly the minimization of the squared distances to the data of ( 2 ) .
the procedure converges because it decreases at each step a cost bounded below by zero .
the estimator computed with this procedure is : @xmath60 the term @xmath61 in equation [ eq : frechet ] is the distance in the shape space between the shapes of @xmath50 and @xmath13 .
thus , we recognize in equation [ eq : frechet ] the frchet mean on the shape space .
the frchet mean is a definition of mean on manifolds @xcite : it is the point that minimizes the squared distances to the data in the shape space .
all in all , one projects the probability distribution function of the @xmath13 s from @xmath14 to @xmath39 and computes its `` expectation '' , in a sense made precise later .
we illustrate the procedure with the examples of the plane and the sphere .
we take @xmath28 , @xmath29 , @xmath62 three objects in @xmath6 in figure [ fig : estimator ] ( a ) and on @xmath9 in figure [ fig : estimator ] ( b ) .
step ( 1 ) is the registration / alignment step .
one filters out the position / parameterization component , i.e. the coordinate on the orbit .
one projects the objects @xmath28 , @xmath29 , @xmath62 in the shape space @xmath24 using the blue arrows .
step ( 2 ) is the computation of the frchet mean of the registered data .
we implemented the generative model and the estimation procedure on the plane and the sphere in shiny applications available online : https://nmiolane.shinyapps.io/shinyplane and https://nmiolane.shinyapps.io/shinysphere .
we invite the reader to look at the web pages and play with the different parameters of the generative model .
figure [ fig : shinyplane1 ] shows screen shots of the applications .
[ fig : shinyplane1 ] our main result is to show that this procedure gives an inconsistent estimate of the template shape @xmath50 of the generative model .
the estimator @xmath63 converges when the number of data goes to infinity .
however it has an asymptotic bias with respect to the parameter @xmath50 it is designed to estimate : @xmath64 $ ] .
this is a vector at the tangent space of @xmath39 at the real parameter @xmath50 .
the vector represents how much one has to shoot from @xmath50 to get the estimated parameter @xmath63 .
it simply writes @xmath65 $ ] for linear spaces .
we could also consider the variance of the estimator .
the variance is defined as @xmath66)^2]$ ] . in the limit of an infinite sample , we have : @xmath67 .
this is why we focus on the asymptotic bias .
we first compute the asymptotic bias for the examples of the plane and the sphere to give the intuition .
the probability distribution function of the @xmath13 s comes from the generative model .
this is a probability distribution on @xmath6 for the plane example , parameterized in polar coordinates @xmath7 like figure [ fig : simple ] .
so we can compute the projected distribution function on the shapes , which are the radii @xmath8 here .
this is done simply by integrating out the distribution on @xmath11 , the position on the circles .
this gives a probability distribution on @xmath37 for the plane example .
we write it @xmath68 .
we remark that @xmath69 does not depend on the probability distribution function on the @xmath70 s of step 2 of the generative model .
we can also compute @xmath71 in the sphere example : we integrate over @xmath72 the probability distribution function on @xmath73 . figure [ fig : pdfs ] ( a ) shows @xmath69 for the plane example , for a template @xmath74 .
we plot it for two different noise levels @xmath75 and @xmath76 .
note that here @xmath69 is the rice distribution .
figure [ fig : pdfs ] ( b ) shows @xmath69 for the sphere example , for a template @xmath77 .
we plot it for different noise levels and @xmath75 and @xmath76 . in both cases
, the x - axis represents the shape space which is @xmath37 for the plane example and @xmath78 $ ] for the sphere example .
the green vertical bar represents the template shape , which is 1 in both cases .
the red vertical bar is the expectation of @xmath69 in each case .
it is @xmath63 , the estimate of @xmath50 we see on these plots that @xmath69 is not centered at the template shape : the green and red bars do not coincide .
@xmath69 is skewed away from 0 in the plane example and away from @xmath36 and @xmath79 in the sphere example .
the skew increases with the noise level @xmath0 .
the difference between the green and red bars is precisely the bias of @xmath63 with respect to @xmath50 .
figure [ fig : bias ] shows the bias of @xmath63 with respect to @xmath50 , as a function of @xmath0 , for the plane ( left ) and the sphere ( right ) .
increasing the noise level @xmath0 takes the estimate @xmath63 away from @xmath50 .
the estimate is repulsed from @xmath36 in the plane example : it goes to @xmath80 when @xmath81 .
it is repulsed from @xmath36 and @xmath79 in the sphere example : it goes to @xmath82 when @xmath83 .
one can show numerically that the bias varies as @xmath84 around @xmath85 in both cases .
this is also observed on the shiny applications @xcite at https://nmiolane.shinyapps.io/shinyplane and https://nmiolane.shinyapps.io/shinysphere .
these examples already show the origin of the asymptotic bias of @xmath63 . _
the bias comes from the curvature of the template s orbit .
_ figure [ fig : curvature ] shows the template s orbit in blue , in ( a ) for the plane and ( b ) for the sphere . in both cases
the black circle represents the level set @xmath0 of the gaussian noise .
the probability of generating an observation @xmath13 outside of the template s shape orbit is bigger than the probability of generating it inside : the grey area in the black circle is bigger than the white area in the white circle .
there will be more registered data that are greater than the template .
their expected will therefore be greater than the template and thus biased .
we prove this in the general case in the next section .
we show the asymptotic bias of @xmath63 in the general case and prove that it comes from the external curvature of the template s orbit . we show it for @xmath50 a principal shape and for a gaussian noise of variance @xmath84 , truncated at @xmath86 .
our results will need the following definitions of curvature .
the _ second fundamental form @xmath87 _ of a submanifold @xmath88 of @xmath14 is defined on @xmath89 by @xmath90 , where @xmath91 denotes the orthogonal projection of covariant derivative @xmath92 onto the normal bundle .
mean curvature vector @xmath93 of @xmath88 _ is defined as : @xmath94 . intuitively , @xmath87 and @xmath93 are measures of extrinsic curvature of @xmath88 in @xmath14 .
for example an hypersphere of radius @xmath95 in @xmath2 has mean curvature vector @xmath96 .
[ th : pdf ] the probability distribution function on the shapes induced by the generative model is : @xmath97 here @xmath98 is the riemannian logarithm @xmath99 $ ] of the shape @xmath100 $ ] at the template shape @xmath50 , @xmath87 is the second fundamental form of the orbit of @xmath98 , @xmath95 is the ricci curvature , @xmath101 are constants independent of @xmath98 and @xmath0 .
the proof is given in appendix [ app : proofpdf ] . the exponential in the expression of @xmath69 belongs to a gaussian distribution centered at @xmath102 .
this is a riemannian gaussian centered at the template shape @xmath50 , because @xmath98 are coordinates at the tangent space of @xmath39 at @xmath50 .
however the whole distribution @xmath69 differs from the gaussian because of the @xmath98-dependent term in the right parenthesis .
this induces a skew of the distribution away from the singular shapes , as observed for the examples in figure [ fig : pdfs ] .
[ th : bias ] the asymptotic bias of the template s shape estimation writes : @xmath103 here @xmath93 is the mean curvature vector of the template shape s orbit and @xmath84 the variance of the noise on the objects . the proof is given in appendix [ app : proofbias ] .
this generalizes the quadratic behavior observed in the examples on figure [ fig : bias ] .
the asymptotic bias has a geometric origin : it comes from the external curvature of the template s orbits , see figure [ fig : curvature ] .
we can vary two parameters in equation [ eq : bias ] : @xmath50 and @xmath0 .
the external curvature of orbits generally increases when @xmath50 is closer to a singularity of the shape space ( see section 1 ) @xcite .
the singular shape of the two landmarks in @xmath6 arises when their distance is 0 . in this case
, the mean curvature vector is @xmath104 : it is inversely proportional to @xmath105 , the radius of the orbit .
@xmath105 is also the distance of @xmath50 to the singularity @xmath36 .
[ [ beyond - y - being - a - principal - shape ] ] beyond @xmath50 being a principal shape + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + our results are valid when the template @xmath50 is a principal shape .
this is a reasonable assumption as the set of principal shapes is dense in the shape space .
what happens when @xmath50 approaches a singularity , i.e. when @xmath50 changes stratum in the stratified space @xmath24 ?
taking the limit @xmath106 in the coefficients of the taylor expansion is not a legal operation .
therefore , we can not conclude on the taylor expansion of the bias for @xmath106 .
indeed , the taylor expansion may even change order for @xmath106 .
we take @xmath107 with the action of @xmath108 and the template @xmath109 : @xmath110 the bias is linear in @xmath0 in this case .
[ [ beyond - sigma-1 ] ] beyond @xmath111 +
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the assumption @xmath111 is reasonable as we hope that the noise on the data is not too large
. nevertheless it would be very interesting to study the asymptotic bias for any @xmath0 , including large noises ( @xmath112 ) .
the distribution over the @xmath13 s in @xmath14 will be spread on the whole manifold @xmath14 .
we can not rely on local computations on @xmath14 ( at the scale of @xmath0 ) anymore .
we have to make global assumptions on the manifold @xmath14 .
the plane example is the canonical example of a flat manifold .
the sphere example is the canonical example of manifold with constant ( positive ) curvature .
the bias as a function of @xmath0 is plotted in figure [ fig : bias ] .
it leads us to the conjecture that the estimate converges towards a barycenter of shape space s singularities when the noise level increases .
singularities have a repulsive action on the estimation of each template s shape .
such repulsive force acts on each estimators . as a result ,
the estimators of the mean shape finds an equilibrium position : the barycenter .
[ [ beyond - one - dirac - in - q - several - templates ] ] beyond one dirac in @xmath24 : several templates + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we have considered so far that there is a unique template shape @xmath50 : the generative model has a dirac distribution at @xmath50 in the shape space .
what happens for other distributions ?
we assume that there are @xmath113 template shapes @xmath114 .
observations are generated in @xmath14 from each template shape @xmath115 with the generative model of section 2 .
our goal is to unveil the structure of the shape distribution , i.e. the @xmath113 template shapes here , given the observations in @xmath14 .
the distributions on shapes projected on the shape space is a mixture of probability density functions of the form of equation [ eq : f ] .
its modes are related to the template shapes .
the k - means algorithm is a very popular method for data clustering .
we study what happens if one uses k - means algorithms on shapes generated with the generative model above . the goal is to cluster the shape data in @xmath113 distinct and significant groups .
one performs a coordinate descent algorithm on the following function : @xmath116 in other words , one minimizes @xmath117 by successively minimizing on the assignment labels @xmath118 s and the cluster s centers @xmath119 s .
given the @xmath118 , minimizing @xmath117 with respect to the @xmath119 s is exactly the simultaneous computation of @xmath113 frchet means in the shape space .
one looks for meaningful well separated clusters ( high inter - clusters dissimilarity ) whose members are close to each other ( high intra - cluster similarity ) . in other words ,
the quality of the clustering is evaluated by the following criterion : @xmath120 which is the dissimilarity between clusters quotiented by the diameter of the clusters . in the absence of singularity in the shape space
, the projected distribution looks like figure [ fig : kmeans ] ( a ) and @xmath121 . the criterion is worse in the presence of singularities .
figure [ fig : kmeans ] illustrates this behavior for the plane example .
we consider any two clusters @xmath122 and call @xmath123 , @xmath124 the estimated centroids .
the criterion @xmath125 writes : @xmath126 even in the best case with correct assignments to the clusters @xmath127 and @xmath128 , the k - means algorithm looses an order of validation when computed on shapes .
[ [ beyond - the - finite - dimensional - case ] ] beyond the finite dimensional case + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + our results are valid when @xmath14 is a finite dimensional manifold and @xmath18 a finite dimensional lie group .
some interesting examples belong to the framework of infinite dimensional manifold with infinite dimensional lie groups .
this is the case for the lddmm framework on images @xcite
. it would be important to extend these results to the infinite dimensional case .
we take @xmath107 with the action of @xmath108 .
we have a analytic expression of @xmath69 in this case @xcite .
figure [ fig : finitedims ] shows the influence of the dimension @xmath129 for the probability distribution functions on the shape space and for the bias .
the bias increases with @xmath129 .
this leads to think that it appears in infinite dimensions as well .
we propose two procedures to correct the asymptotic bias on the template s estimate .
they rely on the bootstrap principle @xcite , more precisely a parametric bootstrap . as such , they are directly applicable to any type of data .
we assume that we know the variance @xmath130 from the experimental setting . the first procedure is called iterative bootstrap .
algorithm [ alg : iterative ] details it .
figure [ fig : iterativebootstrap ] illustrates it on the plane example .
algorithm [ alg : iterative ] starts with the usual template s estimate @xmath131 , see figure [ fig : iterativebootstrap ] ( a ) . at each iteration , we correct @xmath63 with a better approximation of the bias .
first , we generate bootstrap data by using @xmath63 as the template shape of the generative model .
we perform the template s estimation procedure with the frchet mean in the shape space .
this gives an estimate @xmath132 of @xmath133 .
the bias of @xmath134 with respect to @xmath133 is @xmath135 .
it gives an approximation of the bias @xmath136 , see figure [ fig : iterativebootstrap ] ( b ) .
we correct @xmath63 by this approximation of the bias .
this gives a new estimate @xmath137 , see figure [ fig : iterativebootstrap ] ( c ) .
we recall that the bias @xmath136 depends on @xmath50 , see theorem [ th : bias ] .
@xmath137 is closer to the template @xmath50 than @xmath133 .
thus , the next iteration gives a better approximation @xmath138 of @xmath136 .
we correct the initial @xmath63 with this better approximation of the bias , etc .
the procedure is written formally for a general manifold @xmath14 in algorithm [ alg : iterative ] .
[ alg : iterative ] * input : * objects @xmath12 , noise variance @xmath84 + * initialization : * + @xmath139\}_{i=1}^n)$ ] + @xmath140 + * repeat : * + generate bootstrap sample @xmath141 from @xmath142 + @xmath143_i\}_{i=1}^n)$ ] + @xmath144 + @xmath145 + @xmath146 + * until convergence : * @xmath147 + * output : @xmath148 * in algorithm [ alg : iterative ] , @xmath149 denotes the parallel transport from @xmath150 to @xmath151 . for linear spaces , @xmath152 , @xmath153 , @xmath154 .
algorithm [ alg : iterative ] is a fixed - point iteration @xmath155 where : @xmath156 in a linear setting we have simply @xmath157 .
one can show that @xmath158 is a contraction and that @xmath50 , the template shape , is the unique fixed point of @xmath158 ( using the local bijectivity of the riemannian exponential and the injectivity of the estimation procedure ) .
thus the procedure converges to @xmath50 in the case of an infinite number of observations @xmath159 .
figure [ fig : iterativebootstrap_fixedpoint ] illustrates the convergence for the plane example , with a gaussian noise of standard deviation @xmath160 .
the template shape @xmath161 was initially estimated at @xmath162 .
algorithm [ alg : iterative ] corrects the bias .
figures [ fig : iterationsplane ] and [ fig : iterationssphere ] show the iterations of iterative bootstrap for the plane and the sphere example . the second procedure is called the nested bootstrap .
algorithm [ alg : nested ] details it .
figure [ fig : nestedbootstrap ] illustrates it on the plane example .
algorithm [ alg : nested ] starts like algorithm [ alg : iterative ] with @xmath163 , see figure [ fig : nestedbootstrap ] ( a ) .
it also performs a parametric bootstrap with @xmath164 as the template , computes the bootstrap replication @xmath165 and the approximation @xmath166 of @xmath167 , see figure [ fig : iterativebootstrap ] ( b ) .
now algorithm [ alg : nested ] differs from algorithm [ alg : iterative ] .
we want to know how biased is @xmath166 as an estimate of @xmath167 ? this is a valid question as the bias depends on the template @xmath50 , see theorem [ th : bias ] .
we want to estimate this dependence .
we perform a bootstrap , nested in the first one , with @xmath168 as the template .
we compute the estimate @xmath169 and the approximation @xmath170 of @xmath166 , see figure [ fig : iterativebootstrap ] ( c ) .
we observe how far @xmath170 is from @xmath166 .
this gives the blue arrow , which is the bias of @xmath170 as an estimate of @xmath166 , see figure [ fig : iterativebootstrap ] ( d ) .
the blue arrow is an approximation of how far @xmath166 is from @xmath167 .
we correct our estimation of the bias ( in red ) by the blue arrow .
we correct @xmath171 by the bias - corrected estimate of its bias , see figure [ fig : iterativebootstrap ] ( e ) .
[ alg : nested ] * input : * objects @xmath12 , noise variance @xmath84 + * initialization : * + @xmath139\}_{i=1}^n)$ ] + * bootstrap : * + generate bootstrap sample @xmath172 from @xmath173 + @xmath174_i\}_{i=1}^n)$ ] + @xmath175 + * nested bootstrap : * + for each @xmath127 : * generate bootstrap sample @xmath176 from @xmath177 * @xmath178_i\}_{k=1}^n)$ ] @xmath179 + @xmath180 + * output : @xmath137 * one may use the iterative bootstrap or the nested bootstrap depending on the experimental setting .
we illustrate them both on the plane example in figure [ fig : comparison ] .
the advantages of the iterative bootstrap are the following .
it corrects perfectly the bias of @xmath63 in the case of a very large number of observations @xmath181 .
it can be used to experimentally compute the mean curvature vector @xmath93 of each orbit of a group action .
one probes the orbit s curvature by `` feeling it '' with a riemannian gaussian on @xmath14 and projecting on the shape space .
its drawbacks are the following .
it works only with very large @xmath181 .
it is not robust as it uses the generative model several times .
if the generative model is far from being true , then the iterative bootstrap fails .
the advantages of the nested bootstrap are the following .
it is a standard statistical procedure that is more robust with respect to variations of the generative model .
even if generative model is different from the one that we assume , the nested bootstrap performs well .
moreover , it does not need as many data as the iterative bootstrap .
its drawback is that it does not correct perfectly the bias , especially when the noise is important .
these simulations give a rule of thumb for when the bias needs to be corrected .
this is when the noise @xmath0 is comparable to the distance of the template @xmath50 to the singularity .
we perform a simulation using the iterative bootstrap on triangles .
we randomly generate @xmath182 triangles in @xmath6 .
the mean triangle is chosen by taking two coordinates randomly from a uniform distribution on @xmath183 $ ]
. then we add bivariate gaussian noise on each landmark .
these experiments are illustrated in figure [ fig : iterationstriangles ] .
the number of iterations required for the convergence of algorithm 1 with respect to the noise level are shown in figure [ fig : iterationstriangles ] .
we observe the convergence in the three experiments for less than 10 iterations .
now we go to real triangle data .
we have 24 images of rhesus monkeys eyes , acquired with a heidelberg retina tomograph @xcite . for each monkey ,
an experimental glaucoma was introduced in one eye , while the second eye was kept as control .
one seeks a significant difference between the glaucoma and the control eyes . on each image ,
three anatomical landmarks were recorded : @xmath184 for the superior aspect of the retina , @xmath185 for the nose side of the retina , and @xmath186 for the side of the retina closest to the temporal bone of the skull .
the data are matrices @xmath12 where the landmark coordinates form the rows . for the onh example , @xmath14 is the space of @xmath187 landmarks in 3d , @xmath188 and the rotations act isometrically on each object @xmath13 .
* analysis * this simple example illustrates the estimation of the template shape .
we use the following procedure to compute the mean shape for each group .
we initialize @xmath63 with @xmath28 and repeat the following two steps until convergence : @xmath189 figure [ fig : onh ] shows the mean shapes @xmath190 of the control group ( left ) and @xmath191 of the glaucoma group ( right ) in orange , while the initial data are in grey . the difference between the two groups is quantified by the distance between their means : @xmath192 m .
we want to determine if this analysis presents an bias that significantly changes the estimated shape difference between the groups .
we use the nested bootstrap to compute an approximation of the asymptotic bias on each mean shape , for a range of noise s standard deviation in @xmath193 .
the asymptotic bias on the template shape of the glaucoma group is @xmath194 @xmath195 and of the control group is @xmath196 .
the corrected template shape differences are @xmath197 .
in particular , for @xmath198 , we observe that the bias in the template shape are respectively @xmath199 for the healthy group and @xmath200 for the glaucoma group .
this follows the rule - of - thumb : the bias is more important for the healthy group , for which the overall size is smaller than the glaucoma group , for a same noise level .
the bias of the template shape estimate accounts for less than @xmath201 in this case , which is less than @xmath202% of the shapes sizes .
this computation guarantees that this study has not been significantly affected by the bias .
we estimate the impact of the bias on statistics on protein shapes .
a standard hypothesis in biology is that structure ( i.e. shape ) and function of proteins are related .
fundamental research questions about protein shapes include structure prediction - given the protein amino - acid sequence , one tries to predict its structure - and design - given the shape , one tries to predict the sequence needed .
one relies on experimentally determined 3d structures gathered in the protein data base ( pdb ) @xcite .
they contain errors on the protein s atoms coordinates .
average errors range from 0.01 @xmath203 to 1.76 @xmath203 , which is of the magnitude of the length of some covalent bonds .
these values are averaged over the whole protein and in general , the main - chain atoms are better defined than the side - chain atoms or the atoms at the periphery .
this is illustrated on figure where we have plot the b - factor ( related to coordinates errors @xcite ) as a colored map on the atoms for proteins of pdb - codes 1h7w and 4hbb . [ [ proteins - radius - of - gyration ] ] protein s radius of gyration + + + + + + + + + + + + + + + + + + + + + + + + + + + + a biased estimate of a protein shape has consequences for studies on proteins folding .
stability and folding speed of a protein depend on both the estimated shape of the denatured state ( unfolded state ) and of the native state ( folded state ) .
one may study if compact initial states yield to faster folding .
the protein compactness is represented by the protein s radius of gyration , defined as : @xmath204 , where @xmath185 is the number of non - hydrogen atoms , @xmath205 , @xmath206 are resp .
the coordinates of atoms and centers and @xmath207 their mass .
note that we assume ( as it is usually the case ) that all masses @xmath207 for non - hydrogen atoms are equal and that hydrogen atoms have mass @xmath36 .
error on atoms coordinates give a bias on the estimate of the radius of gyration : @xmath208 the radius of protein hjsj ( 85 residues ) is known to be around 10 @xmath203 .
the error on @xmath209 is of @xmath210% with an average error of positions on the atoms of @xmath211 .
it is 8.6% for an error of @xmath212 .
the error will be greater if one consider binding sites at the periphery of the proteins rather than the whole protein .
indeed sites size is smaller and they have less atoms
. one could think about doing clustering on radii of gyration using the k - means algorithm on shapes .
the index @xmath125 of section [ sec : correction ] is : @xmath213 clustering on radii of gyration may lead to a misleading indicator .
@xmath125 indicates that the clustering performs better that it actually does .
[ [ false - positive - probability - in - proteins - motif - detection ] ] false positive probability in protein s motif detection + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the relation between a protein s shape and function is linked to its motifs , which define the supersecondary structure .
motifs have biological properties : for example the helix turn
helix motif @xciteis responsible for the binding of dna within several prokaryotic proteins .
automatic motif detection is another challenge in the study of protein shapes .
we investigate the impact of bias on the false positive probability estimation in motif detection .
let us consider a set @xmath214 of proteins each with @xmath215 atoms .
one is interested in the motifs of @xmath1 atoms that can be detected in the protein s set , where @xmath216 .
we define @xmath0 that represents an allowed error zone . the number of detected motifs increases if : ( i ) one decreases @xmath1 , or ( ii ) one increases @xmath0 , or ( iii ) increases @xmath181 .
thus how many detected motifs actually come from chance , with respect to the parameters @xmath1 , @xmath0 , @xmath181 ?
the false positives probability indicates when one detects truth and when one detects noise .
the usual estimate of the false positive probability is @xmath217 .
here @xmath218 is the volume of the error zone allowed .
@xmath219 is the total volume of the protein @xcite , thus the a ball of radius the radius of gyration .
thus @xmath219 may be biased and overestimated .
the probability of false positive is underestimated .
we consider the example of @xcite .
one tries to find motifs between the tryptophan repressor of escherichia coli ( pdb code 2wrp ) and the cro protein of phage 434 ( pdb code 2cro ) .
these two proteins are known to share the helix - turn - helix motif .
the radius of gyration of 2wrp is @xmath220 , the total volume is : @xmath221 .
we assume an error zone that takes the form of a diagonal covariance matrix with standard deviations @xmath222 .
we get the error zone volume : @xmath223 and the estimation of the false positive probability : @xmath224 .
we find that @xmath225 is underestimated by @xmath226 using the expression of the radius of gyration s bias .
we apply the rule of thumb of section [ sec : correction ] to determine when the bias needs a correction in the computation of a brain template from medical images . here
@xmath14 and @xmath18 will be infinite dimensional .
nevertheless we apply our results to get intuition for this application . in neuroimaging ,
a template is an image representing a reference anatomy .
computing the template is often the first step in medical image processing .
then , the subjects anatomical shapes may be characterized by their spatial deformations _ from the template_. these deformations may serve for ( i ) a statistical analysis of the subject shapes , or ( ii ) for automated segmentation by mapping the template s segmented regions into the subject spaces . in both cases , if the template is not centered among the population , i.e. if it is biased , then the analyzes and conclusions could be biased .
we are interested in highlighting the variables that control the template s bias .
the framework of large deformation diffeomorphic metric mapping ( lddmm ) @xcite embeds the template estimation in our geometric setting .
the lie group of diffeomorphisms acts on the space of images as follows : @xmath227 the isotropy group of @xmath228 writes : @xmath229 .
its lie algebra @xmath230 consists of the infinitesimal transformations whose vector fields are parallel to the level sets of @xmath228 : @xmath231 .
the orbit of @xmath228 is : @xmath232 .
the `` shape space '' is by definition the space of orbits .
two images that are diffeomorphic deformations of one another are in the same orbit .
they correspond to the same point in the shape space .
topology of an image is defined as the image s properties that are invariant by diffeomorphisms .
consequently , the shape space is the space of the images topology , represented by the topology of their level sets .
we get a stratification of the shape space when we gather the orbits by orbit type .
a stratum is more singular than another , if it has higher orbit type , i.e. larger isotropy group .
the manifold @xmath14 has an infinite stratification .
one changes stratum every time there is a change in the topology of an image s level sets .
singular strata are toward simpler topology .
`` principal '' strata are toward a more complicated topology .
indeed , the simpler the topology of the level sets is , the higher is the `` symmetry '' of the image .
thus the larger is its isotropy group .
note that strata with smaller isotropy group ( more detailed topology ) do not represent `` singularities '' from the point of view of a given image and do not influence the bias .
in fact , such strata are at distance 0 : an infinitesimal local change in intensity can create a maximum or minimum , thus complexifying the topology . using the rule - of - thumb of section
[ sec : correction ] , the template s bias depends on its distance @xmath105 to the next singularity , at the scale of @xmath0 the intersubjects variability .
the template is biased in the regions where the difference in intensity between maxima and minima is of the same amplitude as the variability .
the template may converge to pure noise in these regions .
we introduced tools of statistics on manifolds to study the properties of template s shape estimation in medical imaging and computer vision .
we have shown its asymptotic bias by considering the shape space s geometry .
the bias comes from the external curvature of the template s orbit at the scale of the noise on the data .
this provides a geometric interpretation for the bias observed in @xcite .
we investigated the case of several templates and the performance k - mean algorithms on shapes : clusters are less well separated because of each centroid s bias .
the variables controlling the bias are : ( i ) the distance in shape space from the template to a singular shape and ( ii ) the noise s scale .
this gives a rule - of - thumb for determining when the bias is important and needs correction .
we proposed two procedures for correcting the bias : an iterative bootstrap and a nested bootstrap .
these procedures can be applied to any type of shape data : landmarks , curves , images , etc .
they also provide a way to compute the external curvature of an orbit .
our results are exemplified on simulated and real data .
many studies use the template s shape estimation algorithm in molecular biology , medical imaging or computer vision .
their estimations are necessarily biased . but these studies often belong to a regime where the bias is not important ( less than @xmath233 ) .
for example , the bias is important in landmark shapes analyses when the landmarks noise is comparable to the template shape s size .
studies are rarely in this regime .
we have considered shapes belonging to infinite dimensional shape spaces .
our results do not apply to the infinite dimensional case .
we have used them to gain intuition about it .
the bias might be more important in infinite dimensions and need the correction we have suggested .
biblabel[1]#1 . 10 , _ the riemannian geometry of orbit spaces . the metric , geodesics , and integrable systems _ , publ
debrecen , 62 ( 2003 ) . , _ towards a coherent statistical framework for dense deformable template estimation _ , journal of the royal statistical society . , 69 ( 2007 ) ,
329 . , _ estimating the template in the total space with the frchet mean on quotient spaces may have a bias .
_ , proceedings of the fifth international workshop on mathematical foundations of computational anatomy ( mfca15 ) , 2015 , pp . 131142 . ,
_ convergent stochastic expectation maximization algorithm with efficient sampling in high dimension .
application to deformable template model estimation _ , computational statistics & data analysis , 91 ( 2015 ) , pp . 4 19 . , _ the protein data bank _ , nucleic acids res , 28 ( 2000 ) , pp .
235242 .
, _ on the consistency of frchet means in deformable models for curve and image analysis _ , electronic journal of statistics , ( 2011 ) , pp .
10541089 . , _ a deconvolution approach to estimation of a common shape in a shifted curves model _ , ann
, 38 ( 2010 ) , pp .
24222464 . ,
_ the helix - turn - helix dna binding motif .
_ , journal of biological chemistry , 264 ( 1989 ) , pp . 19036 . , _ analyse biomtrique de lanneau pelvien en 3 dimensions
propos de 100 scanners _ ,
revue de chirurgie orthopdique et traumatologique , 100 ( 2014 ) , pp .
s241 . , _ statistical shape analysis _ , john wiley & sons , new york , 1998 . , _
bootstrap methods : another look at the jackknife _ , the annals of statistics , 7 ( 1979 ) , pp . 126 . , _ morphometrics for nonmorphometricians _ , springer , 2012 . , _ brain templates and atlases _ , neuroimage , 62(2 ) ( 2012 ) , pp
. 911922 .
, _ intrinsic shape analysis : geodesic principal component analysis for riemannian manifolds modulo lie group actions . _ , statistica sinica , 20 ( 2010 ) , pp .
1100 . , _ riemannian structures on shape spaces : a framework for statistical inferences _ , in statistics and analysis of shapes , 2006 , pp . 313333 . ,
_ the diffusion of shape _ , advances in applied probability , 9 ( 1977 ) , pp .
428430 .
, _ shape manifolds , procrustean metrics , and complex projective spaces _ , bulletin of the london mathematical society , 16 ( 1984 ) , pp
. 81121 . , _
signal estimation under random time - warpings and nonlinear signal alignment _ , in advances in neural information processing systems 24 , 2011 , pp
. 675683 . , _ the riemannian structure of euclidean shape spaces : a novel environment for statistics _ , the annals of statistics , 21 ( 1993 ) , pp .
12251271 . ,
_ euclidean distance matrix analysis ( edma ) : estimation of mean form and mean form difference _ , mathematical geology , 25 ( 1993 ) , pp .
573602 . . ,
_ mapping the effects of a@xmath234 levels on the longitudinal changes in healthy aging : hierarchical modeling based on stationary velocity fields _ , in proceedings of medical image computing and computer assisted intervention ( miccai ) , vol .
6892 of lncs , springer , 2011 , pp .
663670 . , _ curvature explosion in quotients and applications _ , j. differential geom
. , 85 ( 2010 ) , pp . 117140 . , _ biased estimators on quotient spaces _ , proceedings of the 2nd international of geometric science of information ( gsi2015 ) , ( 2015 ) . , _ nonparametric statistics on manifolds and their applications to object data analysis _ , taylor & francis group , 2016 . , _ toward a generic framework for recognition based on uncertain geometric features _ ,
videre : journal of computer vision research , 1 ( 1998 ) , pp .
5887 . , _ intrinsic statistics on riemannian manifolds : basic tools for geometric measurements _ , journal of mathematical imaging and vision , 25 ( 2006 ) , pp . 127154
, _ a geometric algorithm to find small but highly similar 3d substructures in proteins .
_ , bioinformatics , 14 ( 1998 ) , pp . 516522 . ,
_ riemannian geometry
_ , encyclopaedia of mathem .
sciences , springer , 2001 . , _ easy web applications in r. _ , 2013 .
url : http://www.rstudio.com / shiny/. , _ error estimates of protein structure coordinates and deviations from standard geometry by full - matrix refinement of * @xmath235*b- and * @xmath236*b2-crystallin _ , acta crystallographica section d , 54 ( 1998 ) , pp .
243252 . ,
_ shapes and diffeomorphisms _ , applied mathematical sciences , springer london , limited , 2012 .
here @xmath21 is a point in @xmath14 .
we consider that @xmath21 belongs to a principal orbit @xmath237 .
this will have no impact on the integration because the set of principal orbits is dense in @xmath14 .
we write @xmath238 the projection of @xmath21 in the shape space .
we write the template shape @xmath50 .
we write its estimate @xmath63 .
we take a normal coordinate system centered at the template @xmath50 .
we have the decomposition @xmath239 , where @xmath88 is the orbit of @xmath50 .
we note that @xmath240 , so that @xmath241 are _ not _ the coordinates of @xmath238 in the normal coordinate system at @xmath50 , see figure [ fig : notations ] .
we denote @xmath129 the dimension of @xmath14 , @xmath46 the dimension of the principal orbits and @xmath242 the dimension of the quotient space .
we write coordinates in @xmath24 with indices @xmath243 $ ] and coordinates in an orbit with indices @xmath244 $ ] .
the generative model implies the following riemannian normal distribution on the objects : @xmath245 the distance @xmath246 expressed in the normal coordinate system at @xmath50 is simply @xmath247 . the riemannian measure @xmath248 at @xmath21 in the normal coordinate system at @xmath50 has the taylor expansion : @xmath249 .
we recognize the riemannian curvature tensor @xmath95 .
we truncate the riemannian gaussian : @xmath250 where @xmath184 is the normalization coefficient coming of the univariate truncated gaussian at @xmath251 .
@xmath252 , @xmath253 and @xmath254 refer to geodesic balls of radius @xmath255 in their respective spaces .
we denote @xmath256 and remark that @xmath257 where @xmath242 is the dimension of the quotient space and @xmath46 the dimension of the principal orbits .
we represent @xmath88 in @xmath260 as the graph of a smooth function from @xmath261 to @xmath262 , around @xmath263 by : @xmath264 the local graph @xmath18 is : @xmath265 .
its 0-th and 1-th order derivatives are zero because the graph goes through @xmath238 and is tangent at @xmath266 .
its second order derivative is @xmath267 and represents the best quadratic approximation of the graph .
we need the coordinate of @xmath238 in the ncs at @xmath50 . we parallel transport @xmath270 from @xmath268 to @xmath50 : @xmath271 .
the @xmath36-th order term of the parallel transport at @xmath268 in a ncs at @xmath50 is 0 : @xmath272 this concludes .
we take an ncs at @xmath50 .
we denote @xmath98 the coordinate of @xmath238 in the shape space .
we compute the induced probability distribution @xmath69 on shapes by integrating the distribution on the orbit of @xmath21 out of @xmath273 : @xmath274 the point @xmath21 has coordinates @xmath275 where @xmath276 is the integration coordinate .
@xmath277 the lie group action is isometric so the riemannian metric splits onto @xmath280 at any point .
moreover , the taylor expansion of the metric around @xmath21 still respects the splitting at the third order .
the measure @xmath281 for @xmath276 close enough to @xmath21 is the restriction of this taylor expansion to the orbit .
we have : @xmath282 so that : @xmath283 where @xmath95 is the ricci curvature . for the last term ( 3 ) , we assume again that the graph of the orbit is smooth enough so that it admits a taylor serie at @xmath263 , ie : @xmath286 . by plugging this , we get the moments with the same manipulation as in the proof of theorem 1
. as the moments of order @xmath187 ( and of odd order ) are zero : @xmath287 .
we want to compute the systematic bias .
to this aim , we compute the mean of the distribution of shapes in @xmath24 , in a coordinate system centered at the real @xmath50 .
it writes : @xmath289 the coordinates of @xmath21 in the euclidean coordinate system at @xmath50 are simply @xmath290 .
thus @xmath291 . from the lemma
above : @xmath292 we denote ( 1 ) the first term of the sum and ( 2 ) the second term of the sum .
we compute them independently .
@xmath296 where the last line comes from the definitions of the moments of the truncated gaussian .
we perform a taylor expansion under the integration , in @xmath0 , using the chain rule : @xmath303 again , the terms in @xmath304 integrate to 0 .
we have : @xmath305 the term @xmath306 is precisely the normalization constant of the truncated gaussian @xmath307 , so that : @xmath308 the coefficient @xmath309 comes from the fact that we have truncated the gaussian .
its expression is independent of @xmath0 because we have truncated at a multiple of @xmath0 .
now we show that the second term ( 2 ) is of order @xmath310 .
we assume that the local graph of the orbit is smooth enough so that it has a multivariate taylor serie : @xmath286 .
we have : @xmath312 here the @xmath313 depend on @xmath241 .
we first perform a majoration on each on them by a @xmath314 ( integration on a compact ball ) .
@xmath315 we compute : @xmath316 so that : @xmath317 where we recognize the moment of the truncated gaussian , which are non zero only for even power .
this gives : @xmath318 . | we use tools from geometric statistics to analyze the usual estimation procedure of a template shape .
this applies to shapes from landmarks , curves , surfaces , images etc .
we demonstrate the asymptotic bias of the template shape estimation using the stratified geometry of the shape space .
we give a taylor expansion of the bias with respect to a parameter @xmath0 describing the measurement error on the data .
we propose two bootstrap procedures that quantify the bias and correct it , if needed .
they are applicable for any type of shape data .
we give a rule of thumb to provide intuition on whether the bias has to be corrected .
this exhibits the parameters that control the bias magnitude .
we illustrate our results on simulated and real shape data .
shape , template , quotient space , manifold |
let us consider a familiar case of `` systems '' : the human beings .
human beings are generally considered as the highest peak of biological evolution .
their behavioral and teleological characteristics @xcite set them apart from other system classes @xcite and make them appear to be more `` gifted '' than other beings , e.g. , dogs .
but how do the superior qualities of mankind translate in terms of _ resilience _ ? under stressful or turbulent conditions we know that often a man will result `` better '' than a dog : superior awareness , consciousness , manual and technical dexterity , and reasoning ; advanced ability to reuse experience , learn , develop science , as well as other factors , they all lead to the apparently `` obvious '' conclusion that mankind has a greater ability to tolerate adverse conditions . and though
, it is also quite easy to find counterexamples .
if a threat , e.g. , comes with ultrasonic noise , a dog may perceive the threat and react for instance by running away while a man may stay unaware until too late . or consider the case of miners : inability to perceive toxic gases makes them vulnerable to , e.g. , carbon monoxide and dioxide , methane , and other lethal gases @xcite . a simpler system able to perceive the threat and flee would have more chances to survive .
perception of course is but one of a number of `` systemic features '' that need to be available in order to counterbalance a threat .
so how do we tell whether a system is fit to stand the new conditions characterizing a changing environment ? how do we reason about the quality of resilience ? and , even more importantly , how do we make sure that a system `` stays fit '' if the environment changes ?
the above questions are discussed and , to some extent , addressed in this paper .
our starting point here is the conjecture that _ resilience is no absolute figure _ ; rather , it is _ the result of a match with a deployment environment_. whatever its structure , organization , architecture , capabilities , and resources , a system is only robust as long as its `` provisions '' ( its system characteristics , including the ability to develop knowledge and `` wisdom '' ) match the current environmental conditions .
a second cornerstone of the present discussion is given by the assumption that the interactions between systems and environments can be expressed and reasoned upon by considering the behaviors expressed during those interactions .
in other words , a system - environment fit is the result of the match between the behaviors exercised by a system and those exercised by its environment ( including other systems , the users , etc . )
a third and final assumption is that reasoning about a system s resilience is facilitated by considering the behaviors of those system `` organs '' ( namely , sub - systems ) responsible for the following abilities : 1 .
the ability to perceive change ; 2 .
the ability to ascertain the consequences of change ; 3 .
the ability to plan a line of defense against threats deriving from change ; 4 .
the ability to enact the defense plan being conceived in step 3 ; 5 . and , finally , the ability to treasure up past experience and continuously improve , to some extent , abilities 14 .
as can be clearly seen , the above abilities correspond to the components of the so - called mape - k loop of autonomic computing @xcite .
we shall refer to those abilities as well as the organs that embed them as to the `` systemic features . '' in what follows we first focus in sect .
[ s : sf ] on the concept of behavior and recall the five major classes of behaviors according to rosenblueth , wiener , and bigelow @xcite and boulding @xcite .
we then introduce a system s _ cybernetic class _ by associating each of the systemic features with its own behavior class .
after this , in sect .
[ s : fit ] , we introduce a behavioral formulation of the concepts of supply and system - environment fit as measures of the optimality of a given design with respect to the current environmental conditions .
section [ s : or ] then suggests how proactive and/or social behaviors that would be able to track supply and system - environment fit would pave the way to systems able to self - tune their systemic features in function of the experienced or predicted environmental conditions .
an application of the concepts presented in this work is briefly described in sect .
[ s : ls ] .
our conclusions are finally stated in sect .
[ s : end ] .
as mentioned above , an important attribute towards achieving robustness is given by what we called in sect .
[ s : intro ] as the `` systemic features '' , or the behaviors typical of the system under scrutiny .
such behaviors are the subject of the present section . in what follows we first recall in sect .
[ s : sf : bc ] what are the main behavioral classes .
the main sources here are the classic works by rosenblueth , wiener , and bigelow @xcite and boulding @xcite . in the first work
, classes were identified by the authors by considering the system in isolation . in the second one
boulding introduced an additional class considering the social dimension .
after this , in sect .
[ s : sf : cc ] , we consider an exemplary system ; we identify in it the main system organs responsible for resilience ; and associate behavioral classes to those organs . by doing
so we characterize @xmath0 , namely the `` cybernetic class '' of the system under consideration .
already 71 years ago rosenblueth , wiener , and bigelow @xcite introduced the concept of the `` behavioristic study of natural events '' , namely `` the examination of the output of the object and of the relations of this output to the input '' .
the term `` object '' in the cited paper corresponds to that of `` system '' . in that renowned text
the authors purposely `` omit the specific structure and the intrinsic organization '' of the systems under scrutiny and classify them exclusively on the basis of the quality of the `` change produced in the surroundings by the object '' , namely the system s behavior .
the authors identify in particular four major classes of behaviors : : : : random behavior .
this is an active form of behavior that does not appear to serve a specific purpose or reach a specific state .
a source of electro - magnetic interference exercises random behavior .
: : : purposeful behavior .
this is behavior that serves a purpose and is directed towards a specific goal . quoting the authors , in purposeful behavior we can observe a `` final condition toward which the movement [ of the object ] strives '' .
servo - mechanisms are examples of purposeful behavior . : : : reactive behavior .
this is behavior that `` involve[s ] a continuous feed - back from the goal that modifies and guides the behaving object '' .
examples of this behavior include phototropism , namely the tendency we observe , e.g. , in certain plants , to grow towards the light , and gravitropism , viz . the tendency of plant roots to grow downward .
reactive behaviors require the system to be open @xcite ( able that is to continuously perceive , communicate , and interact with external systems and the environment ) and to embody some form of feedback loop .
: : : proactive behavior .
this is behavior directed towards the extrapolated future state of the goal .
the authors in @xcite classify proactive behavior according to its `` order '' , namely the amount of context variables taken into account in the extrapolation .
kenneth boulding in his classic paper @xcite introduces an additional class : : : : social behaviors .
this class is based on the concept of social organization . quoting the author , in such systems ``
the unit is not perhaps the person the individual human as such but the ` role'that part of the person which is concerned with the organization or situation in question , and it is tempting to define social organizations , or almost any social system , as a set of role tied together with channels of communication . ''
social behaviors may take different forms and be , e.g. , mutualistic , commensalistic , co - evolutive , or co - opetitive @xcite . for more information
we refer the reader to @xcite .
we shall define @xmath1 as a projection map returning , for each of the above behavior classes , an integer in @xmath2 ( @xmath3 , , @xmath4 ) . for any behavior and any set of context figures
@xmath5 , notation will be used to denote that is exercised by considering the context figures in @xmath5 . thus if , for instance , @xmath6 , then refers to a reactive behavior that responds to changes in speed and light . for any behavior and any integer @xmath7
, notation will be used to denote that is exercised by considering @xmath8 context figures , without specifying which ones . as an example , behaviour , with @xmath5 defined as above , identifies an order2 proactive behavior while says in addition that that behavior considers both speed and luminosity to extrapolate the future position of the goal .
we now introduce the concept of partial order among behaviors .
[ d : partial.order ] given any two behaviors @xmath9 and @xmath10 we shall say that @xmath11 if and only if either of the following conditions holds : 1 .
@xmath13 + @xmath14 .
@xmath15 + @xmath16 . whenever two behaviors @xmath9 and @xmath10 are such that @xmath11 , it is possible to define some notion of distance between the two behaviors by considering an arithmetization based on , e.g. , the following factors used as exponents of three different prime numbers : 1 .
2 . @xmath18 .
3 . @xmath19 . in what follows
we shall assume that some metric function , @xmath20 , has been defined .
the behavioral classes recalled in [ s : sf : bc ] may be applied to the five `` systemic features '' introduced in sect .
[ s : intro ] . for any system
@xmath21 we shall refer to the systemic features of @xmath21 through the following 5-tuple : @xmath22 whose components orderly correspond to the abilities introduced in sect .
[ s : intro ] as well as to the stages of mape - k loops @xcite .
system @xmath21 will me omitted when it can be implicitly identified without introducing ambiguity .
for any given system @xmath21 we define as cybernetic class the 5-tuple @xmath23 where , for any @xmath24 , @xmath25 represents the behavior class assigned to systemic feature @xmath26 of @xmath21 , or @xmath27 if @xmath21 does not include @xmath26 altogether . as can be clearly understood , a system s cybernetic class is a qualitative metric that does not provide a full coverage of the systemic characteristics of the system .
as such it should be complemented with quantitative assessments of the quality of service of its system organs namely the sub - systems responsible for hosting its systemic features . in particular for @xmath28 and @xmath29namely , the features corresponding to the abilities of perception and actuation it is useful to complement the notion of behavior with a characterization of the set of context variables that are under the `` sphere of action '' of the corresponding organs . for @xmath28
this means specifying the set of context figures that may be timely perceived by @xmath21 @xcite .
interestingly enough , this concept closely corresponds to that of the powers of representation in leibniz @xcite .
when considering @xmath29 , the sphere of action could be represented by the set of the context figures that may be controlled to a certain extent through system behaviors .
we observe that features @xmath30 and @xmath31 are intrinsically purposeful .
we believe that notation provides a convenient and homogeneous way to express the behavior class and the spheres of action of both @xmath32 and @xmath33 organs .
it is now possible to characterize a system s cybernetic class through notation . as an example , by following the assessments proposed in @xcite , the adaptively redundant data structures described in @xcite have the following cybernetic class @xmath34 while the adaptive @xmath35-version programming system introduced in @xcite is @xmath36 we believe the notion and notation of cybernetic class provide a convenient way to compare qualitatively the systemic features of any two systems with reference to their robustness . as an example , by comparing the above 5-tuples @xmath37 and @xmath38 one may easily realize how the major strength of those two systems lies in their analytic organs , both of which are capable of proactive behaviors ( ) though in a simpler fashion in @xmath37 .
another noteworthy difference is the presence of a knowledge organ in @xmath38 , which indicates that the second system is able to accrue and make use of the past experience in order to improve its action to some extent and exclusively through behaviors .
we conjecture that the action of the knowledge organ in this case corresponds to so - called _ antifragility _
@xcite , namely the ability to `` treasure up '' the past experience so as to improve one s system - environment fit .
what presented in sect . [ s : sf ] allows for a system to be characterized to some extent in terms of its `` systemic features''the provisions that is that play a role when responding to change . as a way to identify the `` quality '' of those provisions in that section we made use of the different behavioral classes as defined in @xcite , and introduced @xmath39 as well as its components .
here we move our attention to a second aspect that , we conjecture , needs to be considered when assessing a system s resilience .
this second aspect tells us how the cybernetic class matches the requirements of dynamically changing environmental conditions . as already anticipated in sect .
[ s : intro ] , in what follows we assume that the evolution of an environment may also be expressed as a behavior . said behavior may be of any of the types listed in sect .
[ s : sf : bc ] and as such it may result in the dynamic variation of a number of `` firing context figures '' .
in fact those figures characterize and , in a sense , set the boundaries of an _ ecoregion _ , namely `` an area defined by its environmental conditions '' @xcite . an environment may be the result of the action of , e.g. , a human being ( a `` user '' ) , or a software managing an ambient , or for instance it may be the result of purposeless ( random ) behavior such as a source of electro - magnetic interference . as a consequence ,
an environment may behave randomly or exhibit a recognizable trend ; in the latter case the variation of its context figures may be such that it allows for tracking or speculation ( extrapolation of future states ) .
moreover , an environment may exhibit the same behavior for a relatively long period of time or it may vary dynamically its character
. we shall refer in what follows to the dynamic evolution of environmental behavior as to an environment s * turbulence*. diagrams such as the one in fig .
[ f : env ] may be used to represent the dynamic evolution of environments .
it is now possible to propose a definition of two indicators for the quality of resilience : the system supply relative to an environment and the system - environment fit .
[ d : supply ] given a system @xmath40 deployed in an environment @xmath33 , characterized respectively by behaviors @xmath41 and @xmath42 ; and given a metric function @xmath20 ; we define as supply at time @xmath43 with respect to @xmath42 the following value : @xmath44 supply can be positive ( oversupply ) , negative ( undersupply ) , or zero ( perfect supply ) .
[ d : fit ] given the same conditions as in definition [ d : supply ] , we define as the system - environment fit at time @xmath43 the function @xmath45 the above definition expresses system - environment fit as a function returning 1 in the case of best fit ; slowly scaling down with oversupply ; and returning @xmath46 in case of undersupply .
it is not the only possible such definition of course : an alternative one is given , for instance , by having @xmath47 instead of @xmath48 .
figure [ f : fitset ] exemplifies a system - environment fit in the case of two behaviors @xmath49 and @xmath50 with @xmath51 .
@xmath33 consists of five context figures identified by integers @xmath52 while @xmath40 consists of context figures @xmath53 .
the system behavior is assumed to be constant ; if @xmath54 this means that the system s perception organ constantly monitors the four figures @xmath53 . on the contrary @xmath50
varies with time .
five time segments are exemplified ( @xmath55 ) during which the following context figures are affected : @xmath56 : : : figures @xmath53 .
@xmath57 : : : figure @xmath58 and figure @xmath59 . @xmath60 : : : figure @xmath59 .
@xmath61 : : : figures @xmath53 .
@xmath62 : : : figures @xmath52
. figures are represented as boxed integers , with an empty box meaning that the figure is not affected by the environment and a filled box meaning the figure is affected .
the behaviour of the environment is constant within a time segment and changes at the next one .
this is shown through the sets at the bottom of fig .
[ f : fitset ] : for each segment @xmath63 the superset is @xmath64 while the subset is @xmath65 , namely @xmath66 .
the relative supply and the system - environment fit also change with the time segments . during @xmath56 and @xmath61
there is perfect supply and best fit : the behavior exercised by the environment is evenly matched by the features of the system . during @xmath57 and @xmath60 the systemic features are more than enough to match the current environmental conditions a case of what we referred to as `` oversupply '' .
correspondingly , fit is rather low .
in @xmath62 we have the opposite situation : the systemic features for instance , pertaining to a perception organ are insufficient to become aware of all the changes produced by the environment .
in particular here changes connected with figure 5 go undetected .
this is a case of `` undersupply '' , corresponding to the `` worst possible '' system - environment fit .
the two functions introduced in sect . [ s : fit ] , @xmath48 and @xmath67 , may be interpreted as measures of the optimality of a given design with respect to the current environmental conditions .
whenever those conditions allow it and a partial order `` @xmath68 '' exists for the behaviors at play , then it is possible to consider system behaviors of the following forms : 1 .
@xmath69 , with @xmath5 including figures @xmath48 and @xmath67.[e : opt.pro ] + such behavior , when exercised by system organs for analysis , planning , and knowledge management , translates in the possibility to become aware and speculate on the possible future robustness requirements . if this is coupled with the possibility to revise one s system organs by enabling or disabling , e.g. , the ability to perceive certain context figures depending on the extrapolated future environmental conditions , then a system could proactively improve its own system - environment fit .
2 . @xmath70 , with @xmath5 including figures @xmath48 and @xmath67.[e : opt.soc ] + analysis , planning , and knowledge management behaviors of this type aim at artificially augmenting or reducing the system features by establishing / disestablishing collaborative relationships as exemplified in the `` canary - in - the - coal - mine '' scenario of @xcite . as we did in the paper
just cited we propose to call behaviors such as [ e : opt.pro ] ) and [ e : opt.soc ] ) as * auto - resilient*. finally , we remark how the formulation of system - environment fit presented in this work may also be tailored so as to include overheads and costs .
littlesister @xcite is an icon project financed by the iminds research institute and the flemish government agency for innovation by science and technology ( iwt ) .
the project aims to deliver a low - cost telemonitoring @xcite solution for home care and is to run until the end of year 2014 .
littlesister adopts a connectionist approach in which the collective action of an interconnected network of simple units @xcite ( battery - powered mouse sensors ) replaces the adoption of more powerful and expensive complex devices ( smart cameras ) . in order for this approach to be effective
the mentioned collective action is to guarantee that an optimal trade - off between energy efficiency , performance , and safety is dynamically sustained .
we plan to express this optimal trade - off in terms of a system - environment fit
. obviously the formulation of the littlesister system - environment fit will be considerably more complex than the one introduced in the present work .
a key role will be played in particular by the littlesister awareness organ , which will be used to determine the level of criticality of the current situation and set an operative mode ranging from `` energy - saving - first '' to `` safety - first '' .
this operative mode will be included in the set of context figures of the social behavior @xmath70 of littlesister s sensors .
depending on the requirements expressed by the current operative mode and other context figures , the system - environment fit will vary , which will translate in a variable selection and number of sensors to be activated .
the goal we aim to reach is being able to sustain at the same time both maximum safety and minimum energy expenditure .
the questions we have posed in sect . [ s : intro ] have been answered , to some extent , by defining a conceptual framework for their discussion .
the nature of our framework is behavioral and `` sits on the shoulders '' of the work carried out in the first half of last century by `` giants '' such as bogdanov , wiener , von bertalanffy , boulding , and several others in turn based on the intriguingly modern ideas of `` elder giants '' such as leibniz @xcite and aristotle @xcite . within our framework
we have introduced a behavioral formulation of the concepts of supply and system - environment fit as measures of the optimality of a system with respect to the current conditions of the environment in which the system is deployed .
moreover , we have suggested how complex abilities such as auto - resilience and antifragility may be expressed in terms of behaviors able to track supply and fit measures and evolve the systemic features in function of the hypothesized future environmental conditions .
practical application of the concepts in this article has been briefly discussed by considering a strategy for optimizing the collective behavior of the mouse sensors used in project littlesister . as can be clearly understood , our work is far from being exhaustive or complete .
in particular discussing context figures without referring to a `` range '' , or sphere of action , makes it difficult to compare behaviors such as auditory perception in animals .
our future work will include extending our conceptual framework accordingly .
another direction we intend to take is the application of our concepts towards the design of antifragile computing systems ; the reader may refer to @xcite for a few preliminary ideas about this .
i would like to express my gratitude to alan carter for helping me with the pictures in this paper .
this work was partially supported by iminds interdisciplinary institute for technology , a research institute funded by the flemish government as well as by the flemish government agency for innovation by science and technology ( iwt ) .
the iminds littlesister project is a project co - funded by iminds with project support of iwt .
partners involved in the project are universiteit antwerpen , universiteit gent , vrije universiteit brussel , xetal , christelijke mutualiteit vzw , niko projects , jf oceans bvba , and sbd nv .
a. rosenblueth , n. wiener , and j. bigelow , `` behavior , purpose and teleology , '' _ philosophy of science _ ,
vol . 10 , no . 1 , pp . 1824 , 1943 .
[ online ] .
available : http://www.journals.uchicago.edu/doi/abs/10.1086/286788 v. de florio , `` preliminary contributions towards auto - resilience , '' in _ proceedings of the 5th international workshop on software engineering for resilient systems ( serene 2013 ) , lecture notes in computer science 8166_.1em plus 0.5em minus 0.4emkiev , ukraine : springer , october 2013 , pp .
141155 .
f. heylighen , `` basic concepts of the systems approach , '' in _ principia cybernetica web _ , f. heylighen , c. joslyn , and v. turchin , eds.1em plus 0.5em minus 0.4emprincipia cybernetica , brussels , 1998 . [ online ] .
available : http://pespmc1.vub.ac.be/sysappr.html r. adner and r. kapoor , `` value creation in innovation ecosystems : how the structure of technological interdependence affects firm performance in new technology generations , '' _ strategic management journal _
31 , pp . 306333 , 2010 .
k. gdel , `` on formally undecidable propositions of principia mathematica and related systems , ''
november 2000 , translation and adaptation by martin hirzel of gdel s `` ber formal unentscheidbare stze der principia mathematica und verwandter systeme '' .
v. de florio , `` on the role of perception and apperception in ubiquitous and pervasive environments , '' in _ proceedings of the 3rd workshop on service discovery and composition in ubiquitous and pervasive environments ( supe12 ) _ , august 2012 .
[ online ] .
available : http://www.sciencedirect.com/science/article/pii/s1877050912005297 g. leibniz and l. strickland , _ the shorter leibniz texts : a collection of new translations _ , ser .
continuum impacts.1em plus 0.5em minus 0.4emcontinuum , 2006 .
[ online ] .
available : http://books.google.be/books?id=ofocy3xj8nkc v. de florio , `` on the constituent attributes of software and organisational resilience , '' _ interdisciplinary science reviews _ , vol . 38 , no . 2 , june 2013 .
[ online ] .
available : http://www.ingentaconnect.com / content / maney / isr/2013/00000038/00000002/% art00005[http://www.ingentaconnect.com / content / maney / isr/2013/00000038/00000002/% art00005 ] v. de florio and c. blondia , `` reflective and refractive variables : a model for effective and maintainable adaptive - and - dependable software , '' in _ proc . of the 33rd euromicro conference on software engineering and advanced applications ( seaa 2007 ) _ , lbeck , germany , august 2007 .
j. buys , v. de florio , and c. blondia , `` towards context - aware adaptive fault tolerance in soa applications , '' in _ proceedings of the 5th acm international conference on distributed event - based systems ( debs)_. 1em plus 0.5em minus 0.4emassociation for computing machinery , inc .
( acm ) , 2011 , pp . 6374 .
j. buys , v. de florio , and c. blondia , `` towards parsimonious resource allocation in context - aware n - version programming , '' in _ proceedings of the 7th iet system safety conference_.1em plus 0.5em minus 0.4em the institute of engineering and technology , 2012 . v. de florio , `` antifragility = elasticity + resilience + machine learning : models and algorithms for open system fidelity , '' _ arxiv e - prints _
2014 , submitted for publication in the proceedings of the 1st international workshop `` from dependable to resilient , from resilient to antifragile ambients and systems '' ( antifragile 2014 ) .
s. meystre , `` the current state of telemonitoring : a comment on the literature , '' _ telemed j e health _ , vol .
11 , no . 1 ,
pp . 6369 , 2005 .
[ online ] .
available : http://www.ncbi.nlm.nih.gov/pubmed/15785222 | already 71 years ago rosenblueth , wiener , and bigelow introduced the concept of the `` behavioristic study of natural events '' and proposed a classification of systems according to the quality of the behaviors they are able to exercise . in this paper
we consider the problem of the resilience of a system when deployed in a changing environment , which we tackle by considering the behaviors both the system organs and the environment mutually exercise .
we then introduce a partial order and a metric space for those behaviors , and we use them to define a behavioral interpretation of the concept of system - environment fit . moreover we suggest that behaviors based on the extrapolation of future environmental requirements would allow systems to proactively improve their own system - environment fit and optimally evolve their resilience . finally we describe how we plan to express a complex optimization strategy in terms of the concepts introduced in this paper . |
statistical correlations between the brightness of the chromosphere and corona strongly suggest that these two regions are powered by a similar driver @xcite . yet , despite some suggestions that much of the non - thermal energy driving coronal heating is deposited in the chromosphere @xcite , most models of coronal loops treat the chromosphere as a passive mass reservoir that merely reacts to energy release at coronal heights and the associated changes in the coronal thermal conductive flux or deposition of non - thermal energy @xcite .
the differing role of the chromosphere in these models implies that observations of how the corona is filled with plasma from the chromosphere hold significant diagnostic potential for distinguishing between various coronal heating models @xcite .
yet , it has been very difficult to establish how the dynamics and energetics of chromosphere and corona are coupled @xcite .
chromospheric spicules have long been considered a candidate for such coupling by providing mass to the corona and/or solar wind @xcite .
these jet - like features , with lifetimes of order 3 - 10 minutes , propel cool matter upwards to coronal heights with speeds of 20 - 30 km / s @xcite . a significant fraction of these are likely caused by magnetoacoustic shocks on magnetic flux concentrations @xcite .
spicules are observed at temperatures from 5,000 to 500,000 k and are estimated to carry upward a mass flux 100 times larger than that of the solar wind @xcite .
however , no signatures of these ejecta at coronal temperatures have been reported previously @xcite . consequently , a direct role of spicules in the coronal mass / energy balance has been dismissed as unlikely @xcite .
however , the high spatio - temporal resolution of the solar optical telescope ( sot ; * ? ? ?
* ) onboard _ hinode _
@xcite has revealed a previously unrecognized second type of spicules @xcite that is a prime candidate for establishing a link between corona and chromosphere .
these `` type ii '' spicules have much shorter lifetimes ( 10 - 100 s ) , are more violent ( upflows of order 50 - 150 km / s ) , and often fade rapidly from the chromospheric h 3968 sot passband , which suggests rapid heating and ionization of singly to at least doubly ionized calcium .
the formation of these ubiquitous type ii spicules has been attributed to magnetic reconnection @xcite . to study the thermal evolution of these spicules , we use simultaneous observations of h emission ( with sot ) and euv spectral lines formed in the corona at temperatures of several million k from the extreme - ultraviolet imaging spectrometer ( eis ; * ? ? ?
we use spectra from sumer @xcite on soho @xcite to show that the process we describe is ubiquitous and occurs in active regions , quiet sun and coronal holes .
the observations and our novel analysis techniques are described in robs , while
obs covers the results of our analysis . in
discuss we discuss the impact of our results on the existing coronal heating hypotheses .
we focus on plage regions in active regions noaa 10976 and 10977 observed near disk center on 2 and 5 december 2007 by hinode / sot and eis . by observing at disk center
we reduce the ambiguities introduced by line - of - sight superposition at the limb ( where spicules are most readily identified ) , which has plagued such studies in the past @xcite .
sot obtained a timeseries of images using the h 3968 filter ( with 0.5s exposure and 8s cadence ) that covered a 25@xmath040@xmath1 area of plage .
simultaneous eis observations of the same region started with a region - wide scan with coarse horizontal ( 5@xmath1 ) stepping with 60s exposures for a wide variety of spectral lines .
this was followed by much finer ( 1@xmath1 stepping ) , rapid ( 60s exposure ) coronal rasters with eis over a 5@xmath1x80@xmath1 sliver of plage . to connect the chromospheric and coronal dynamics we need to coalign the coronal spectral rasters with the high cadence chromospheric imaging .
this is not straightforward .
the sot timeseries are coaligned onboard _
hinode _ using a correlation tracker .
the eis spectra are pointed independently from sot , and undergo significant jitter caused by the spacecraft , and thermal flexing within eis .
we remove the spacecraft jitter from the eis pointing by using the xrt_jitter idl routine ( from solarsoft ) . to correct for the effect of thermal flexing on eis pointing , we use several steps .
first , we perform additional coalignment of the timeseries using cross - correlation .
this removes drifts caused by solar evolution within the small field of view of the sot correlation tracker .
we use chromospheric and coronal images from _ trace _ @xcite to connect the eis and sot data and determine the eis pointing as a function of time .
this pointing information is used to calculate synthetic rasters of upper chromospheric activity ( uca ) derived from the h timeseries .
the latter is dominated by slowly evolving ( on timescales @xmath2s ) photospheric and low - chromospheric contributions because of the wide bandpass ( fwhm=2.2 ) of the sot h filter . to isolate the relatively faint signature ( from the ca h core ) of the upper - chromospheric type - ii spicules in the h timeseries , we exploit their highly dynamic nature : they evolve on much shorter timescales ( @xmath3s ) @xcite than the photospheric and low - chromospheric contributions .
we thus perform , for each pixel , temporal high - pass fourier filtering on the timeseries with a cutoff of 18 mhz .
the resulting timeseries of ( the absolute value of the ) high frequency signal is a good proxy for uca because it removes the photospheric and low - chromospheric features ( online movie 1 ) . to study the high temperature response
, we analyze eis spectra of moss regions which constitute the transition region ( tr ) at the footpoints of active region loops @xcite .
we find that the bright core of the coronal spectral lines are typically doppler shifted by less than 10 - 20 km / s and note that the doppler shift and intensities of the coronal profiles show no spatial correlation between upper - chromospheric and coronal activity @xcite .
however , @xcite recently discovered that there is often a faint blue - shifted component in addition to the bright core of the line , indicating coronal upflows of order 50 - 100 km / s . to isolate these upflows from the bright core of the spectral line more clearly
, we calculate maps of the blue - red ( b - r ) asymmetry in the coronal spectral lines ( fig .
[ fig1 ] ) and find that the upflows occur preferentially in moss footpoint regions . to calculate the b - r asymmetry of spectral lines ,
we first fit a single gaussian to the line profile @xmath4 , and determine the line centroid @xmath5 of the gaussian fit . the b - r asymmetry for an offset @xmath6 from the line centroid is then given by : @xmath7 with @xmath8 the wavelength range over which the b - r asymmetry is determined . to determine the propagated errors on the b - r asymmetry measure , we use estimates for the error on the intensity using the eis software ( including photon noise and other errors ) . in our maps we only show locations where the b - r asymmetry is larger than twice the estimated propagated error .
we apply the same method to sumer rasters of quiet sun and coronal hole on 6-nov-1999 , after calibrating the data using the techniques described by @xcite .
we calculate b - r asymmetry maps for eis rasters of 274 ( formed at 2 mk under equilibrium conditions ) for velocities arond 80 km / s and 120 km / s ( fig . [ fig2]b , c ) .
both b - r asymmetry maps show no obvious correlation with the intensity of the core of the line ( fig . [ fig2]a ) from which they are derived .
in contrast , we find that the sites of faint upflows at coronal temperatures ( fig .
[ fig2]b , c ) are well correlated with upper - chromospheric activity ( fig .
[ fig2]d , online movie 2 ) .
the correspondence between the b - r and uca maps is surprisingly good , given several complicating factors .
first , despite our careful coalignment , there remains uncertainty of order 2 in the eis / sot coalignment .
in addition , the correlation between the chromospheric signal and coronal blueshifts associated with spicules depends greatly on the viewing angle between the line - of - sight and the direction of the upward flows ( fig . 5 ) .
when looking down at loop footpoints , we observe the chromospheric jet as a h brightening ( because of projection ) , associated with a faint , strongly blue - shifted component in the coronal line .
when the line - of - sight is not aligned with the upflows , the spicules will appear as jets in the chromospheric line , but we no longer see blue - shifts in the coronal line since the line - of - sight velocity is reduced , and the spicular emission measure is too small compared to the dominant coronal emission . the wide range of angles between upflows ( i.e. , magnetic field ) and line - of - sight thus naturally leads to an imperfect correlation between chromospheric brightness and coronal upflows .
it is also possible that some of the uca sites that do not match coronal upflow events do have coronal counter parts , but at much higher or lower temperatures than the line we observe .
we also can not expect a one - to - one correlation between the upper chromospheric brightness and the brightness in the coronal asymmetry maps , because we do not know the exact formation and heating mechanism of the spicules .
finally , some of the coronal upflow events could be nanoflares that are driven by energy deposition in the corona , i.e. , without significant chromospheric brightness enhancement .
the relationship between these chromospheric events and coronal upflows is elucidated by our finding that the velocities of the latter in two lines ( fig .
[ fig3 ] ) are similar to the upward velocities for type ii spicules seen at the limb ( from * ? ? ?
* ) , both peaking between 50 and 100km / s .
we find that these upflows are pervasive and universal : uv lines observed with sumer show faint upflowing components with velocities of the same magnitude ( 50 - 100km / s ) that are well correlated across the tr and low corona ( 100,000 to 600,000 k ) for a whole range of magnetic field configurations ( active region plage , and network in quiet sun and coronal holes , with the velocities in the latter two somewhat lower than in plage ) .
this is illustrated with spectroheliograms of 1548 and 770 from for a coronal hole region near disk center ( fig .
[ fig4 ] ) .
we find a clear difference in intensity ( reduced ) and doppler velocity ( blueward ) as well as doppler velocity ( blueward ) in the coronal hole compared to the quiet sun .
these differences can be understood in terms of a picture where the bulk of the material reaching the formation temperature of is injected into the solar wind @xcite .
we also find that the average profile in the network in both quiet sun and coronal hole shows significant asymmetry towards the blue for velocities of 50 - 100 km / s for both and lines .
the b - r asymmetry maps show that faint upflows at 50 - 100 km / s are well correlated for a lower tr ( ) and a coronal ( ) line and preferentially occur in and around the network .
the spatial patterning and correlation of these upflows does not change as one crosses the boundary between quiet sun and coronal hole .
this is surprising , since the magnetic field configuration in the corona is dramatically different .
our analysis of the asymmetry of spectral lines is very sensitive to the presence of blends in the wings of the lines , which can skew the asymmetry maps towards the red or blue so that these are no longer indicative of flows .
for that reason , we have avoided lines that have strong blends .
we have extensively studied the influence of blends on the lines used , and find that none of them change our conclusions ( see a follow - up paper for a detailed discussion ) .
we point out that the upflow maps for different lines ( at 264 and 274 ) are very similar , and that we observe similar upflows for a wide range of temperatures ( 1548 , 770 , 264 and 274 ) .
we estimate that these spicule - associated upflows play a significant role in replenishing the corona with hot mass .
to play such a role in active regions , these hot upflows must provide enough mass flux @xmath9 to replenish the corona which loses mass by downflows ( after cooling ) at a rate of @xmath10 . here , @xmath11 is the density of the coronal mass propelled upward and heated in a spicule , @xmath12 the upward velocity of the spicule , @xmath13 the number of spicules that occur at any location over the spicule lifetime @xmath14 , @xmath15 the density in coronal loops , @xmath16 the coronal scale height , @xmath17 the time for coronal plasma to cool to chromospheric temperatures , and @xmath18 the ratio of cross sections of the upper corona and the plage region below .
our observations in coronal lines also show that the faint blue - shifted spicular emission measure @xmath19 ( with @xmath20 the fraction of total spicular coronal density at a temperature @xmath21 ) can only be a fraction @xmath22 of that of the dominant emission @xmath23 at the core of the lines , which is emitted by the slowly cooling plasma in previously filled loops .
we thus require that @xmath24 and @xmath25 , and find that @xmath26 , and @xmath27 .
for realistic estimates of @xmath28 , @xmath29 km , @xmath30s @xcite , @xmath31 ( from analysis of eis spectra , fig .
[ fig1]c , d ) , @xmath32s and @xmath33km / s @xcite we find @xmath34 and @xmath35 .
observations only cover temperatures around 2 mk , and most likely reveal only a fraction of the coronal component of spicules , which may range in temperature from 1 to 10mk .
if we make the reasonable assumption that only a fraction @xmath36 of the spicular coronal mass flux is observed in , we find , for coronal densities @xmath37@xmath38 , a total density of the coronal component of spicules of order @xmath39@xmath38 .
this implies that these spicules can fill the corona with hot plasma even if only between 1 and 5 % of the chromospheric spicule densities reach coronal temperatures @xcite .
we can also estimate how many spicules @xmath40 are required at any time in a plage region of diameter @xmath41 : @xmath42 with @xmath43 the diameter of a spicule . for @xmath44 km and @xmath45 km ,
@xmath40 is of order 80 - 320 , implying a spicule number density of 0.7 - 3 spicules per square arcsecond of plage .
this density is compatible with sot observations , which implies that these spicules can play a significant role in filling coronal loops .
individual spicular events have energies of the order of nanoflares ( @xmath46 erg ) and must recur every @xmath47 seconds at the same location ( i.e , 250 -1000 s for the above estimates ) to replenish the coronal mass .
the energy flux carried into the corona by these hot spicules is of order @xmath48 erg @xmath49 s@xmath50 with the above assumptions and if we assume that all spicules reach 2 mk temperatures .
this is of the same order of magnitude as the energy flux required to heat the active region corona @xcite .
the spatio - temporal correlation between upper chromospheric activity ( associated with jets ) and faint upflows at several mk , and the ubiquity of faint and correlated upflows with velocities of order 50 - 100 km / s for temperatures ranging from 10,000 to 2 mk for a variety of magnetic field configurations ( active region , quiet sun and coronal hole ) poses significant challenges to our current understanding of the mass cycling of the corona .
the properties of these events do not seem compatible with those predicted by coronal loop models that are driven only by nanoflares at coronal heights .
such models predict high - speed upflows at very high temperatures of several million degrees , with either lower velocities or no detectable upflows at lower temperatures ( @xmath511 million k ) @xcite .
more generally , our observations of similar upflow velocities for a wide range of temperatures from 10,000 k to 2 million k are a challenge to any model in which upflows are driven by transient overpressure at chromospheric heights ( e.g. , by deposition of non - thermal energy from electron beams generated in the corona ; * ? ? ?
* ) because those lead to increasing velocities with temperature .
our finding that these upflows do not change qualitatively or quantitatively between quiet sun and coronal holes strongly suggests that they are driven from below .
this is because models in which energy is deposited at coronal heights are dependent on conductive flux and/or reconnection at tangential discontinuities , which are expected to have significantly different properties in open versus closed field environments @xcite .
such models also do not provide a natural explanation for why the chromospheric counterparts show upward extrusions and can be seen in lines from neutral elements such as mg i b 5172 ( sot ) .
nanoflares at coronal heights may be important for coronal heating , but they can not explain our observations .
our analysis suggests a scenario ( fig .
[ fig5 ] ) where a significant , potentially dominant , part of the heating of the coronal plasma occurs at chromospheric heights , in association with chromospheric jets ( similar to previous suggestions , e.g. , * ? ? ?
a significant fraction of the plasma propelled upward in these chromospheric jets or type ii spicules is heated to coronal temperatures , providing the corona with hot plasma .
the coronal emission of individual spicules is faint compared to the dominant emission , which originates in previously filled loops that are slowly cooling .
previous efforts to establish a connection between chromosphere and corona have focused on properties of the core of coronal lines .
the weak emission of the coronal upflows associated with spicules , which provide the direct link to the chromosphere , helps explain why these efforts have failed in the past ( e.g. , * ? ? ?
conceptually , our findings are compatible with results from numerical simulations @xcite that suggest that most of the heating in the solar atmosphere occurs below heights of 2 mm , i.e. , at chromospheric heights
. preliminary results of more recent simulations suggest a scenario in which magnetic reconnection in the chromosphere causes upward jets of plasma at a range of temperatures owing to magnetic lorentz forces even as the plasma is heated to tr and coronal temperatures in the process .
swm acknowledges support from nsf grant atm-0925177 and nasa grants nng06gc89 g and nnx08ah45 g .
bdp is supported by nasa grants nnm07aa01c ( hinode ) , nng06gg79 g and nnx08ah45 g .
_ hinode _ is a japanese mission developed and launched by isas / jaxa , with naoj as a domestic partner and nasa and stfc ( uk ) as international partners . | the sun s corona is millions of degrees hotter than its 5,000 k photosphere .
this heating enigma is typically addressed by invoking the deposition at coronal heights of non - thermal energy generated by the interplay between convection and magnetic field near the photosphere .
however , it remains unclear how and where coronal heating occurs and how the corona is filled with hot plasma .
we show that energy deposition at coronal heights can not be the only source of coronal heating , by revealing a significant coronal mass supply mechanism that is driven from below , in the chromosphere .
we quantify the asymmetry of spectral lines observed with _
hinode _ and _ soho _ and identify faint but ubiquitous upflows with velocities that are similar ( 50 - 100 km / s ) across a wide range of magnetic field configurations and for temperatures from 100,000 to several million degrees .
these upflows are spatio - temporally correlated with and have similar upward velocities as recently discovered , cool ( 10,000 k ) chromospheric jets or ( type ii ) spicules .
we find these upflows to be pervasive and universal .
order of magnitude estimates constrained by conservation of mass and observed emission measures indicate that the mass supplied by these spicules can play a significant role in supplying the corona with hot plasma .
the properties of these events are incompatible with coronal loop models that only include nanoflares at coronal heights .
our results suggest that a significant part of the heating and energizing of the corona occurs at chromospheric heights , in association with chromospheric jets . |
in bayesian optimization @xcite ( bo ) , we wish to optimize a derivative - free expensive - to - evaluate function @xmath0 with feasible domain @xmath1 , @xmath2 with as few function evaluations as possible . in this paper , we assume that membership in the domain @xmath3 is easy to evaluate and we can evaluate @xmath0 only at points in @xmath3 .
we assume that evaluations of @xmath0 are either noise - free , or have additive independent normally distributed noise .
we consider the parallel setting , in which we perform more than one simultaneous evaluation of @xmath0 .
bo typically puts a gaussian process prior distribution on the function @xmath0 , updating this prior distribution with each new observation of @xmath0 , and choosing the next point or points to evaluate by maximizing an acquisition function that quantifies the benefit of evaluating the objective as a function of where it is evaluated . in comparison with other global optimization algorithms ,
bo often finds `` near optimal '' function values with fewer evaluations @xcite . as a consequence ,
bo is useful when function evaluation is time - consuming , such as when training and testing complex machine learning algorithms ( e.g. deep neural networks ) or tuning algorithms on large - scale dataset ( e.g. imagenet ) @xcite .
recently , bo has become popular in machine learning as it is highly effective in tuning hyperparameters of machine learning algorithms @xcite .
most previous work in bo assumes that we evaluate the objective function sequentially @xcite , though a few recent papers have considered parallel evaluations @xcite .
while in practice , we can often evaluate several different choices in parallel , such as multiple machines can simultaneously train the machine learning algorithm with different sets of hyperparameters . in this paper , we assume that we can access @xmath4 evaluations simultaneously at each iteration .
then we develop a new parallel acquisition function to guide where to evaluate next based on the decision - theoretical analysis . * our contributions . * we propose a novel batch bo method which measures the information gain of evaluating @xmath5 points via a new acquisition function , the parallel knowledge gradient ( @xmath6 ) .
this method is derived using a decision - theoretic analysis that chooses the set of points to evaluate next that is optimal in the average - case with respect to the posterior when there is only one batch of points remaining . naively maximizing @xmath6
would be extremely computationally intensive , especially when @xmath5 is large , and so , in this paper , we develop a method based on infinitesimal perturbation analysis ( ipa ) @xcite to evaluate @xmath6 s gradient efficiently , allowing its efficient optimization . in our experiments on both synthetic functions and
tuning practical machine learning algorithms , @xmath6 consistently finds better function values than other parallel bo algorithms , such as parallel ei @xcite , batch ucb @xcite and parallel ucb with exploration @xcite .
@xmath6 provides especially large value when function evaluations are noisy .
the code in this paper is available at https://github.com/wujian16/qkg .
the rest of the paper is organized as follows .
section [ sect : related ] reviews related work .
section [ sect : gaussian ] gives background on gaussian processes and defines notation used later .
section [ sect : qkg ] proposes our new acquisition function @xmath6 for batch bo .
section [ sect : practice ] provides our computationally efficient approach to maximizing @xmath6 .
section [ sect : numerical ] presents the empirical performance of @xmath6 and several benchmarks on synthetic functions and real problems .
finally , section [ sect : conclusion ] concludes the paper .
[ sect : related ] within the past several years , the machine learning community has revisited bo @xcite due to its huge success in tuning hyperparameters of complex machine learning algorithms .
bo algorithms consist of two components : a statistical model describing the function and an acquisition function guiding evaluations . in practice ,
gaussian process ( gp ) @xcite is the mostly widely used statistical model due to its flexibility and tractability .
much of the literature in bo focuses on designing good acquisition functions that reach optima with as few evaluations as possible . maximizing this acquisition function
usually provides a single point to evaluate next , with common acquisition functions for sequential bayesian optimization including probability of improvement ( pi)@xcite , expected improvement ( ei ) @xcite , upper confidence bound ( ucb ) @xcite , entropy search ( es ) @xcite , and knowledge gradient ( kg ) @xcite .
recently , a few papers have extended bo to the parallel setting , aiming to choose a batch of points to evaluate next in each iteration , rather than just a single point .
@xcite suggests parallelizing ei by iteratively constructing a batch , in each iteration adding the point with maximal single - evaluation ei averaged over the posterior distribution of previously selected points .
@xcite also proposes an algorithm called constant liar " , which iteratively constructs a batch of points to sample by maximizing single - evaluation while pretending that points previously added to the batch have already returned values .
there are also work extending ucb to the parallel setting .
@xcite proposes the gp - bucb policy , which selects points sequentially by a ucb criterion until filling the batch .
each time one point is selected , the algorithm updates the kernel function while keeping the mean function fixed .
@xcite proposes an algorithm combining ucb with pure exploration , called gp - ucb - pe . in this algorithm , the first point is selected according to a ucb criterion
; then the remaining points are selected to encourage the diversity of the batch .
these two algorithms extending ucb do not require monte carlo sampling , making them fast and scalable .
however , ucb criteria are usually designed to minimize cumulative regret rather than immediate regret , causing these methods to underperform in bo , where we wish to minimize simple regret .
the parallel methods above construct the batch of points in an iterative greedy fashion , optimizing some single - evaluation acquisition function while holding the other points in the batch fixed .
the acquisition function we propose considers the batch of points collectively , and we choose the batch to jointly optimize this acquisition function . other recent papers that value points collectively include @xcite which optimizes the parallel ei by a closed - form formula , @xcite , in which gradient - based methods are proposed to jointly optimize a parallel ei criterion , and @xcite , which proposes a parallel version of the es algorithm and uses monte carlo sampling to optimize the parallel es acquisition function .
we compare against methods from a number of these previous papers in our numerical experiments , and demonstrate that we provide an improvement , especially in problems with noisy evaluations .
our method is also closely related to the knowledge gradient ( kg ) method @xcite for the non - batch ( sequential ) setting , which chooses the bayes - optimal point to evaluate if only one iteration is left @xcite , and the final solution that we choose is not restricted to be one of the points we evaluate .
( expected improvement is bayes - optimal if the solution is restricted to be one of the points we evaluate . )
we go beyond this previous work in two aspects .
first , we generalize to the parallel setting .
second , while the sequential setting allows evaluating the kg acquisition function exactly , evaluation requires monte carlo in the parallel setting , and so we develop more sophisticated computational techniques to optimize our acquisition function .
recently , @xcite studies a nested batch knowledge gradient policy . however , they optimize over a finite discrete feasible set , where the gradient of kg does not exist . as a result ,
their computation of kg is much less efficient than ours .
moreover , they focus on a nesting structure from materials science not present in our setting .
in this section , we state our prior on @xmath0 , briefly discuss well known results about gaussian processes ( gp ) , and introduce notation used later .
we put a gaussian process prior over the function @xmath7 , which is specified by its mean function @xmath8 and kernel function @xmath9 .
we assume either exact or independent normally distributed measurement errors , i.e. the evaluation @xmath10 at point @xmath11 satisfies @xmath12 where @xmath13 is a known function describing the variance of the measurement errors .
if @xmath14 is not known , we can also estimate it as we do in section [ sect : numerical ] . supposing we have measured @xmath0 at @xmath15 points @xmath16 and obtained corresponding measurements @xmath17 , we can then combine these observed function values with our prior to obtain a posterior distribution on @xmath0 .
this posterior distribution is still a gaussian process with the mean function @xmath18 and the kernel function @xmath19 as follows @xmath20
in this section , we propose a novel parallel bayesian optimization algorithm by generalizing the concept of the knowledge gradient from @xcite to the parallel setting .
the knowledge gradient policy in @xcite for discrete @xmath3 chooses the next sampling decision by maximizing the expected incremental value of a measurement , without assuming ( as expected improvement does ) that the point returned as the optimum must be a previously sampled point .
we now show how to compute this expected incremental value of an additional iteration in the parallel setting .
suppose that we have observed @xmath15 function values .
if we were to stop measuring now , @xmath21 would be the minimum of the predictor of the gp .
if instead we took one more batch of samples , @xmath22 would be the minimum of the predictor of the gp .
the difference between these quantities , @xmath23 , is the increment in expected solution quality ( given the posterior after @xmath24 samples ) that results from the additional batch of samples .
this increment in solution quality is random given the posterior after @xmath15 samples , because @xmath22 is itself a random vector due to its dependence on the outcome of the samples .
we can compute the probability distribution of this difference ( with more details given below ) , and the @xmath6 algorithm values the sampling decision @xmath25 according to its expected value , which we call the parallel knowledge gradient factor , and indicate it using the notation @xmath6 .
formally , we define the @xmath6 factor for a set of candidate points to sample @xmath26 as @xmath27 , \label{eq : multikg}\ ] ] where @xmath28 : = \mathbb{e } \left[\cdot |{\bm{x}^{(1:n ) } } , { y^{(1:n ) } } \right]$ ] is the expectation taken with respect to the posterior distribution after @xmath15 evaluations .
then we choose to evaluate the next batch of @xmath5 points that maximizes the parallel knowledge gradient , @xmath29 by construction , the parallel knowledge gradient policy is bayes - optimal for minimizing the minimum of the predictor of the gp if only one decision is remaining .
the @xmath6 algorithm will reduce to the parallel ei algorithm if function evaluations are noise - free and the final recommendation is restricted to the previous sampling decisions . because under the two conditions above , the increment in expected solution quality will become @xmath30 which is exactly the parallel ei acquisition function .
however , computing @xmath6 and its gradient is very expensive .
we will address the computational issues in section [ sect : practice ] .
the full description of the @xmath6 algorithm is summarized as follows .
the number of initial stage samples @xmath31 , and the number of main stage sampling iterations @xmath32 .
initial stage : draw @xmath31 initial samples from a latin hypercube design in @xmath3 , @xmath33 for @xmath34 .
main stange : solve ( [ eqn : innerkg ] ) , i.e. get @xmath35 sample these points @xmath36 , re - train the hyperparameters of the gp by mle , and update the posterior distribution of @xmath0 .
in this section , we provide the strategy to maximize @xmath6 by a gradient - based optimizer .
in section [ sect : value ] and section [ sect : gradient ] , we describe how to compute @xmath6 and its gradient when @xmath3 is finite in ( [ eq : multikg ] ) .
section [ sect : discrete ] describes an effective way to discretize @xmath3 in ( [ eq : multikg ] ) .
the readers should note that there are two @xmath3s here , one is in ( [ eq : multikg ] ) which is used to compute the @xmath6 factor given a sampling decision @xmath26 .
the other is the feasible domain in ( [ eqn : innerkg ] ) ( @xmath38 ) that we optimize over .
we are discretizing the first @xmath3 .
following @xcite , we express @xmath39 as @xmath40 because @xmath41 is normally distributed with zero mean and covariance matrix @xmath42 with respect to the posterior after @xmath15 observations , we can rewrite @xmath39 as @xmath43 where @xmath44 is a standard @xmath5-dimensional normal random vector , and @xmath45 where @xmath46 is the cholesky factor of the covariance matrix @xmath42 .
now we can compute the @xmath6 factor using monte carlo sampling when @xmath3 is finite : we can sample @xmath44 , compute ( [ eqn : alternative ] ) , then plug in ( [ eq : multikg ] ) , repeat many times and take average . in this section
, we propose an unbiased estimator of the gradient of @xmath6 using ipa when @xmath3 is finite .
accessing a stochastic gradient makes optimization much easier . by ( [ eqn : alternative ] )
, we express @xmath6 as @xmath47 where @xmath48 . under the condition that @xmath49 and @xmath50 are continuously differentiable
, one can show that ( please see the details in the supplementary materials ) @xmath51 where @xmath52 is the @xmath53th dimension of the @xmath54th point in @xmath26 . by the formula of @xmath55 , @xmath56 where @xmath57 , @xmath58 , and @xmath59 now we can sample many times and take average to estimate the gradient of @xmath6 via ( [ eqn : qkg_grad ] ) .
this technique is called infinitesimal perturbation analysis ( ipa ) in gradient estimation @xcite . since we can estimate the gradient of @xmath6 efficiently when @xmath3 is finite
, we will apply some standard gradient - based optimization algorithms , such as multi - start stochastic gradient ascent to maximize @xmath6 .
we have specified how to maximize @xmath6 when @xmath3 is finite in ( [ eq : multikg ] ) , but usually @xmath3 is infinite . in this case , we will discretize @xmath3 to approximate @xmath6 , and then maximize over the approximate @xmath6 .
the discretization itself is an interesting research topic @xcite . in this paper , the discrete set @xmath60
is not chosen statically , but evolves over time : specifically , we suggest drawing @xmath61 samples from the global optima of the posterior distribution of the gaussian process ( please refer to @xcite for a description of this technique ) .
this sample set , denoted by @xmath62 , is then extended by the locations of previously sampled points @xmath63 and the set of candidate points @xmath26 .
then ( [ eq : multikg ] ) can be restated as @xmath64,\end{aligned}\ ] ] where @xmath65 . for the experimental evaluation
we recompute @xmath62 in every iteration after updating the posterior of the gaussian process .
we conduct experiments in two different settings : the noise - free setting and the noisy setting . in both settings ,
we test the algorithms on well - known synthetic functions chosen from @xcite and practical problems . following previous literature @xcite
, we use a constant mean prior and the ard mat@xmath66rn @xmath67 kernel . in the noisy setting , we assume that @xmath68 is constant across the domain @xmath3 , and we estimate it together with other hyperparameters in the gp using maximum likelihood estimation ( mle ) .
we set @xmath69 to discretize the domain following the strategy in section [ sect : discrete ] . in general
, the @xmath6 algorithm performs as well or better than state - of - art benchmark algorithms on both synthetic and real problems .
it performs especially well in the noisy setting . before describing the details of the empirical results , we highlight the implementation details of our method and the open - source implementations of the benchmark methods .
our implementation inherits the open - source implementation of parallel ei from the ` metrics optimization engine ` @xcite , which is fully implemented in ` c++ ` with a python interface .
we reuse their gp regression and gp hyperparameter fitting methods and implement the @xmath6 method in ` c++ ` . besides comparing to parallel ei in @xcite
, we also compare our method to a well - known heuristic parallel ei implemented in ` spearmint ` @xcite , the parallel ucb algorithm ( gp - bucb ) and parallel ucb with pure exploration ( gp - ucb - pe ) both implemented in ` gpoptimization ` @xcite . in this section ,
we focus our attention on the noise - free setting , in which we can evaluate the objective exactly .
we show that parallel knowledge gradient outperforms or is competitive with state - of - art benchmarks on several well - known test functions and tuning practical machine learning algorithms .
first , we test our algorithm along with the benchmarks on 4 well - known synthetic test functions : branin2 on the domain @xmath70 ^ 2 $ ] , rosenbrock3 on the domain @xmath71 ^ 3 $ ] , ackley5 on the domain @xmath71 ^ 5 $ ] , and hartmann6 on the domain @xmath72 ^ 6 $ ] .
we initiate our algorithms by randomly sampling @xmath73 points from a latin hypercube design , where @xmath74 is the dimension of the problem .
figure [ fig : syn ] reports the mean and the standard deviation of the base 10 logarithm of the immediate regret by running @xmath75 random initializations with batch size @xmath76 .
the results show that @xmath6 is significantly better on rosenbrock3 , ackley5 and hartmann6 , and is slightly worse than the best of the other benchmarks on branin2 . especially on rosenbrock3 and ackley5
, @xmath6 makes dramatic progress in early iterations .
[ fig : syn ] in this section , we test the algorithms on two practical problems : tuning logistic regression on the mnist dataset and tuning cnn on the cifar10 dataset .
we set the batch size to @xmath76 .
first , we tune logistic regression on the mnist dataset .
this task is to classify handwritten digits from images , and is a @xmath77-class classification problem .
we train logistic regression on a training set with @xmath78 instances with a given set of hyperparameters and test it on a test set with @xmath79 instances .
we tune @xmath80 hyperparameters : mini batch size from @xmath77 to @xmath81 , training iterations from @xmath75 to @xmath79 , the @xmath82 regularization parameter from @xmath83 to @xmath84 , and learning rate from @xmath83 to @xmath84 .
we report the mean and standard deviation of the test error for @xmath85 independent runs . from the results
, one can see that both algorithms are making progress at the initial stage while @xmath6 can maintain this progress for longer and results in a better algorithm configuration in general .
, width=604,height=226 ] [ fig : uncertain ] in the second experiment , we tune a cnn on cifar10 dataset .
this is also a @xmath77-class classification problem .
we train the cnn on the @xmath86 training data with certain hyperparameters and test it on the test set with @xmath79 instances . for the network architecture , we choose the one in ` tensorflow ` tutorial .
it consists of @xmath87 convolutional layers , @xmath87 fully connected layers , and on top of them is a softmax layer for final classification .
we tune totally @xmath88 hyperparameters : the mini batch size from @xmath77 to @xmath89 , training epoch from @xmath84 to @xmath77 , the @xmath82 regularization parameter from @xmath83 to @xmath84 , learning rate from @xmath83 to @xmath84 , the kernel size from @xmath87 to @xmath77 , the number of channels in convolutional layers from @xmath77 to @xmath89 , the number of hidden units in fully connected layers from @xmath75 to @xmath89 , and the dropout rate from @xmath83 to @xmath84 .
we report the mean and standard deviation of the test error for @xmath90 independent runs . in this example , the @xmath6 is making better ( more aggressive ) progress than parallel ei even in the initial stage and maintain this advantage to the end .
this architecture has been carefully tuned by the human expert , and achieve a test error around @xmath91 , and our automatic algorithm improves it to around @xmath92 . in this section , we study problems with noisy function evaluations .
our results show that the performance gains over benchmark algorithms from @xmath6 evident in the noise - free setting are even larger in the noisy setting .
we test on the same @xmath80 synthetic functions from the noise - free setting , and add independent gaussian noise with standard deviation @xmath93 to the function evaluation . the algorithms are not given this standard deviation , and must learn it from data .
[ fig : syn ] the results in figure [ fig : uncertain ] show that @xmath6 is consistently better than or at least competitive with all competing methods .
also observe that the performance advantage of @xmath6 is larger than for noise - free problems .
testing on a large test set such as imagenet is slow , especially when we must test many times for different hyperparameters . to speed up hyperparameter tuning , we may instead test the algorithm on a subset of the testing data to approximate the test error on the full set .
we study the performance of our algorithm and benchmarks in this scenario , focusing on tuning logistic regression on mnist .
we train logistic regression on the full training set of @xmath94 , but we test the algorithm by testing on @xmath95 randomly selected samples from the test set , which provides a noisy approximation of the test error on the full test set . ,
width=264,height=302 ] [ fig : uncertain ] we report the mean and standard deviation of the test error on the full set using the hyperparameters recommended by each parallel bo algorithm for @xmath85 independent runs .
the result shows that @xmath6 is better than both versions of parallel ei , and its final test error is close to the noise - free test error ( which is substantially more expensive to obtain ) .
as we saw with synthetic test functions , @xmath6 s performance advantage in the noisy setting is wider than in the noise - free setting .
the authors were partially supported by nsf career cmmi-1254298 , nsf cmmi-1536895 , nsf iis-1247696 , afosr fa9550 - 12 - 1 - 0200 , afosr fa9550 - 15 - 1 - 0038 , and afosr fa9550 - 16 - 1 - 0046 .
in this paper , we introduce a novel batch bayesian optimization method @xmath6 , derived from a decision - theoretical perspective , and develop a computational method to implement it efficiently .
we show that @xmath6 outperforms or is competitive with the state - of - art benchmark algorithms on several synthetic functions and in tuning practical machine learning algorithms . * supplementary material *
the ( [ eqn : innerkg ] ) corresponds to the synchronous @xmath6 optimization , in which we wait for all @xmath5 points from our previous batch to finish before searching for a new batch of @xmath5 points .
however , in some applications , we may wish to generate a new batch of points to evaluate next while @xmath96 points are still being evaluated , before we have their values .
this is common in training machine learning algorithms , where different machine learning models do not necessarily finish at the same time .
@xmath29 we can generalize ( [ eqn : innerkg ] ) to the asynchronous @xmath6 optimization .
given that @xmath97 points are still under evaluation , now we would like to recommend a batch of @xmath5 points to evaluate . as we did for the synchronous @xmath6 optimization above
, now we estimate the @xmath6 of the combined @xmath98 points only with respect to the @xmath5 points that we need to recommend
. then we proceed the same way via gradient - based algorithms .
next , we compare @xmath6 at different levels of parallelism against the fully sequential kg algorithm .
we test the algorithms with different batch sizes on two noisy synthetic functions branin2 and hartmann6 , whose standard deviation of the noise is @xmath99 . from the results ,
our parallel knowledge gradient method does provide a speed - up as @xmath5 goes up .
[ fig : syn ]
recall that in section 5 of the main document , we have expressed the @xmath6 factor as follows , @xmath100 where the expectation is taken over @xmath101 and @xmath102 the main purpose of this section is to prove the following proposition .
when @xmath3 is finite , under the condition that @xmath49 and @xmath50 are continuous differentiable , @xmath103 where @xmath104 , @xmath105 , @xmath52 is the @xmath53th dimension of the @xmath54th point in @xmath26 and @xmath106 . without loss of generality
, we assume that ( 1 ) @xmath54 and @xmath53 are fixed in advance and ( 2 ) @xmath107^d$ ] , we would like to prove that ( [ eqn : qkg_grad_interchange ] ) is correct . before proceeding ,
we define one more notation @xmath108 where @xmath26 equals to @xmath109 component - wise except for @xmath52 . to prove it
, we cite theorem 1 in @xcite , which requires three conditions to make ( [ eqn : qkg_grad_interchange ] ) valid : there exists an open neighborhood @xmath110 of @xmath111 where @xmath111 is the @xmath53th dimension of @xmath54th point in @xmath109 such that ( i ) @xmath112 is continuous in @xmath113 for any fixed @xmath3 and @xmath44 , ( ii ) @xmath112 is differentiable except on a denumerable set in @xmath113 for any given @xmath3 and @xmath44 , ( iii ) the derivative of @xmath112 ( when it exists ) is uniformly bounded by @xmath114 for all @xmath115 , and the expectation of @xmath114 is finite . under the condition that the mean function @xmath49 and the kernel function @xmath50 are continuous differentiable
, we see that for any given @xmath116 , @xmath117 is continuous differentiable in @xmath26 by the result that the multiplication , the inverse ( when the inverse exists ) and the cholesky operators @xcite preserve continuous differentiability .
when @xmath118 is finite , we see that @xmath119 is continuous in @xmath26 .
then @xmath112 is also continuous in @xmath52 by the definition of the function @xmath112 . by the expression that @xmath120 , if both @xmath121 and @xmath122 are unique , then @xmath112 is differentiable at @xmath52 .
we define @xmath123 to be the set that @xmath112 is not differentiable , then we see that @xmath124 where @xmath125 .
@xmath126 depend on @xmath52 if @xmath127 ( @xmath128 ) where @xmath129 is the @xmath54th point of @xmath26 . as @xmath3 is finite ,
we only need to show that @xmath130 and @xmath131 is denumerable . defining @xmath132 on @xmath113
, one can see that @xmath133 is continuous differentiable on @xmath113 .
we would like to show that @xmath134 is denumerable . to prove it
, we will show that @xmath135 contains only isolated points .
then one can use a theorem in real analysis : any set of isolated points in @xmath136 is denumerable ( see the proof of statement 4.2.25 on page 165 in @xcite ) . to prove that @xmath135 only contains isolated points
, we use the definition of an isolated point : @xmath137 is an isolated point of @xmath135 if and only if @xmath138 is not a limit point of @xmath135 .
we will prove by contradiction , suppose that @xmath137 is a limit point of @xmath135 , then it means that there exists a sequence of points @xmath139 all belong to @xmath135 such that @xmath140 .
however , by the definition of derivative and @xmath141 , @xmath142 , a contradiction .
so we conclude that @xmath135 only contains isolated points , so is denumerable .
defining @xmath143 on @xmath113 , @xmath144 is also continuous differentiable on @xmath113 , then one can similarly prove that @xmath145 is denumerable . recall that from section 5 of the main document , @xmath146 where @xmath57 , @xmath147 , and @xmath148 we can calculate the @xmath149 as follows @xmath150 using the fact that @xmath49 is continuously differentiable and @xmath3 is compact , then @xmath149 is bounded by some @xmath151 . by the result that @xmath152 is continous ,
it is bounded by a vector @xmath153 as @xmath3 is compact .
then @xmath154 where @xmath155 . and @xmath156 .
in this section , we will prove that sga converges to a stationary point .
we follow the same idea of proving the theorem 2 in @xcite .
first , it requires the step size @xmath157 satisfying @xmath158 as @xmath159 , @xmath160 and @xmath161 .
second , it requires the second moment of the gradient estimator is finite . in the above section 1.3 ,
we have show that @xmath162 , then @xmath163 . | in many applications of black - box optimization , one can evaluate multiple points simultaneously , e.g. when evaluating the performances of several different neural networks in a parallel computing environment . in this paper
, we develop a novel batch bayesian optimization algorithm the parallel knowledge gradient method . by construction
, this method provides the one - step bayes optimal batch of points to sample .
we provide an efficient strategy for computing this bayes - optimal batch of points , and we demonstrate that the parallel knowledge gradient method finds global optima significantly faster than previous batch bayesian optimization algorithms on both synthetic test functions and when tuning hyperparameters of practical machine learning algorithms , especially when function evaluations are noisy . |
as the most important driver of severe space weather , coronal mass ejections ( cmes ) and their geoeffectiveness have been studied intensively .
previous statistical studies have shown that not all the front - side halo cmes are geoeffective ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and not all non - recurrent geomagnetic storms can be tracked back to a cme ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these phenomena may cause some failed predictions of the geoeffectiveness of cmes .
the recent notable event exhibiting such a failure was on 2015 march 15 when a fast cme originated from the west hemisphere .
space weather prediction center ( swpc ) of noaa initially forecasted that the cme would at most cause a very minor geomagnetic disturbance labeled as g1 , a scale used by swpc to measure the intensity of geomagnetic storms .
however , the cme produced the largest geomagnetic storm so far , at g4 level with the provisional @xmath7 value of @xmath8 nt , in the current solar cycle 24 @xcite .
the major geomagnetic storm was called the 2015 st .
patrick s day " event as its main phase and peak occurred on march 17 , and the surprising cme was selected as a campaign event by international study of earth - affecting solar transients ( isest ) , a program under scostep , and also by coupling , energetics and dynamics of atmospheric regions ( cedar ) .
such an unexpected phenomenon naturally raises the first question for the forecasting of the geoeffectiveness of a cme , i.e. , whether or not a cme will hit the earth even though we know the source location and initial kinematic properties of the cme . a full understanding of the propagation trajectory of a cme from the sun to 1 au is the key to this question .
of course , it is not the only factor determining the geoeffectiveness of a cme .
the magnetic field strength and the orientation of the cme flux rope , which directly affect the strength and duration of the interval of the south - component of the magnetic field , is also important for determining its geoeffectiveness .
it has been well accepted that the cme s trajectory can be deflected in the corona ( within a few tens of solar radii , @xmath9 ) .
@xcite illustrated that such deflections can be classified into three types : asymmetrical expansion , non - radial ejection and deflected propagation . in a statistical sense , cmes tend to be deflected toward the equator during solar minimum ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) or deflected away from coronal holes ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the physics behind these deflections is that the gradient of the magnetic energy density may cause the cme to move toward the place where the magnetic energy density reaches the minimum , usually the location of the heliospheric current sheet @xcite .
although the cme s deflection in the corona could be tens of degrees and may change the geoeffectiveness of a cme , it still can be monitored by coronagraphs ( e.g. , * ? ? ?
thus , the possible deflection of a cme in interplanetary space rather than the deflection in the corona is one of the major sources of uncertainty in the prediction of the cme impact at the earth .
the possibility of the cme deflection in interplanetary space was first proposed by @xcite .
they suggested that , different from the deflection in the corona , the cme s trajectory in interplanetary space could be deflected due to the velocity difference between the cme and the ambient solar wind . for a fast cme ,
the solar wind plasma and interplanetary magnetic field will be piled up from the west and ahead of it , leading to a net deflection force toward the east ; for a slow cme , the picture is the opposite . a kinematic model ( called dips , _ deflection in interplanetary space _ ,
hereafter ) was therefore developed @xcite .
such deflections are thought to be gradual and much slower than that in the corona , but the total amount of the deflection angle is comparable to that in the corona as it takes place over a much longer distance .
such evidence can also be found in previous studies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
one of the most comprehensive analysis of the cme s trajectory in interplanetary space was done by @xcite for a slow cme , which was proven to experience a westward deflection all the way from the corona to 1 au with a total deflection angle of more than 20 degrees . for the 2015 march 15 cme
, there were no stereo ( solar terrestrial relations observatory , * ? ? ? * ) data as the twin spacecraft were behind the sun and not taking images .
all the information of the cme came from the remote - sensing data provided by the solar and heliospheric observatory ( soho ) and the in - situ data by wind ( or ace ) spacecraft at 1 au .
the interplanetary space between the corona to 1 au thus had an observational gap , and therefore the propagation of the cme from the sun to 1 au is unclear . in this paper
, we try to recover the kinematic evolution of the cme from the limited observations and fill the gap with the aid of models .
we particularly focus on the trajectory of the cme to demonstrate how the cme behavior in interplanetary space favors its strong geoeffectiveness .
the fast cme of interest first appeared in the field of view ( fov ) of lasco ( large angle and spectrometric coronagraph , * ? ? ? * ) c2 camera on board the soho on march 15 at about 01:36 ut , and left the fov of lasco / c3 , which monitors the corona within 30 @xmath9 , around 09 ut .
it was a partial halo cme with most material ejected toward the west as shown in figure [ fg_gcs]a . by examining the solar euv images , e.g. ,
figure [ fg_source]a and [ fg_source]b , taken by the atmospheric imaging assembly ( aia , * ? ? ?
* ) on board the solar dynamics observatory ( sdo ) , we can identify that the source region of the cme is active region ( ar ) 12297 on the west hemisphere of the sun with a large coronal hole ( ch ) to the south - east of it .
the eruption took place around the location of w35s15 with filament - like material moving toward the south - west direction .
meanwhile , two bright ribbons can be seen in the aia 1600 passband ( fig.[fg_source]c ) , suggesting a typical eruptive flare .
the two daily @xmath10 images taken by kanzelhoehe observatory before and after the eruption ( fig.[fg_source]d ) did show a disappearance of a segment of a thick filament near the ar .
however , as will be discussed below , there was another slow cme perhaps originating from the same ar between the times of the two @xmath10 images .
thus , the association of the disappeared filament segment in the @xmath10 images to the cme is unclear .
the aia euv images suggest a clear filament eruption during the fast cme , and therefore it is very likely that the disappeared filament is associated with this cme . considering that the filament is a good tracer of the cme flux rope
, one may estimate that the tilt angle , i.e. , the angle between the main axis of the cme flux rope and the solar equator , is initially within the range of @xmath11 .
it should be noted that the coronal conditions during the fast cme traveling through the lasco fovs were complicated .
the movie ( see supplementary material ) made from the soho / lasco images ( or see the example image fig.[fg_gcs]a ) reveals that ( 1 ) there was a structure running faster than the fast cme along the north - west direction , and ( 2 ) there was a slow cme , which was launched earlier on march 14 , propagating toward the south .
the running structure appeared in the lasco fov at the same time as the fast cme appeared .
it not only ran faster than but also looked fainter than the fast cme .
this running structure might be the shock wave driven by the fast cme , or an independent magnetic structure that erupted from the sun .
we incline to the latter because the running structure was much faster than the fast cme and its shape was much narrow than and different from the cme s front .
however , near the eruption of the fast cme , we can not find any other notable eruption signatures on the visible solar disk .
although stealth cmes exist , they tend to be slow ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
thus , the fast running structure was probably a real ejection from the backside of the sun .
since the structure propagated faster than and ahead of the cme of interest , we do not consider any possible interaction between them . for the preceding slow cme , there are two possibilities of its source locations ( due to the lack of stereo observations ) .
one was suggested by gopalswamy and yashiro at the isest workshop in october , 2015 .
they thought that it was a backside event because the pre - existing streamer disturbed by the slow cme was moving to the south pole , suggesting that its location was on the backside .
the other possibility is that it was a front - side cme .
there was a notable eruptive signature in the same ar 12297 around 12:00 ut on march 14 . the time and location match well with the appearance and the speed of the slow cme in the lasco / c2 fov .
if the first possibility was true , there should be no interaction between the slow cme and the fast cme of interest .
but if the second possibility was true , the two cmes might have interacted with each other . with the aid of foward modeling ,
e.g. , the gcs model @xcite , we then analyze the kinematics of the two cmes as well as this possible interaction .
since soho / lasco provides only one angle of view , the gcs fitting suffers from a larger uncertainty than that when the stereo data are available .
thus , we reduce the degrees of freedom during the fitting by setting three free parameters , the tilt angle , aspect ratio and angular width , to be constant , and only vary the other three free parameters , the longitude , latitude and the height . thanks to a sufficient number of images in the time sequence , these free parameters of the gcs model can still be roughly constrained by trial and error .
the uncertainties in these parameters are estimated by following the method of @xcite , i.e. , by decreasing the goodness - of - fit between the leading edge of the cme and the model by 10% . since the leading edge of the cme in all the images
is determined manually by hand - clicks , which may increase the errors , the uncertainties inferred by the above method are underestimated . for the fast cme ,
the best value of the tilt angle is about @xmath12 , the aspect ratio about 0.66 , and the angular width about @xmath13 ( edge - on ) or @xmath14 ( face - on ) . for the preceding slow cme , they are @xmath15 , 0.47 , @xmath16 and @xmath17 , respectively .
the uncertainties in the tilt angle , aspect ratio and the angular width are about @xmath18 , 0.12 and @xmath19 .
these uncertainties are very large , and thus these fitting values are only used for reference . for the other three time - dependent parameters , the fitting results are shown in figure [ fg_gcs]c[fg_gcs]e .
the uncertainty in the longitude is about @xmath3 , that in the latitude is better , about @xmath20 , and that in the height is less than one solar radius .
the two cmes almost propagated along the same longitude , which is around @xmath21 , but in latitude , the two cmes were separated by about @xmath22 . considering the angular width of the two cmes , they might marginally interact with each other if the preceding slow cme was a front - side event . according to the height - time plot shown in figure [ fg_gcs]e
, the two cmes traveled through the lasco fovs at a speed of 817 and 251 km s@xmath4 , respectively .
the leading edge of the fast cme caught up with the leading edge of the preceding slow one around 05:20 ut . thus , the possible interaction should start around 03:00 ut and end before the fast cme left the fov of lasco / c3 .
however , neither considerable acceleration of the preceding slow cme nor deceleration of the fast cme can be found in the height - time profiles .
an interesting phenomenon is that the preceding slow cme was systematically deflected toward lower latitude , and the fast cme slightly toward higher latitude . due to the significant uncertainties
, we can not conclude that the two cmes interacted based on such small deflections .
nevertheless , we can conclude that the two cmes at most experienced a very weak interaction , which has little influence on the kinematics of the fast cme in interplanetary space .
note , there is also a great possibility that the two cmes did not interact at all , because the preceding one might come from the backside of the solar disk .
the inferred kinematic evolution of the two cmes in the corona suggest that the fast cme should be able to encounter the earth but the slow cme is unlikely to encounter the earth . since the fast cme may drive a shock at front
, what we fitted with gcs model is probably not the leading edge of its flux rope but the driven shock .
the recent work by @xcite suggested that the longitudinal extent of the flux rope carried by a cme is typically about @xmath23 , smaller than its driven shock if any .
thus , the angular width of the fast cme derived from the gcs model is probably overestimated .
if this were the case , the fast cme s flux rope might just graze the earth with a weaker geoeffectiveness .
two days later , a fast forward shock followed by a magnetic cloud ( mc ) was recorded by the wind spacecraft @xcite as shown in figure [ fg_wind ] .
the shock arrived at 1 au at 04:00 ut on march 17 .
its driver , the mc , occurred between 13:05 and 23:20 ut , and is characterised by the reduced fluctuation in the magnetic field , the large and smooth rotation of the magnetic field direction , evident bi - directional suprathermal electron beams and the low temperature and proton @xmath24 . according to the classification proposed by @xcite and @xcite , the mc is an esw - type flux rope . a fast stream , probably from the ch ( as seen in fig.[fg_source]a )
, was catching up with the mc , causing an interaction region during 23:20 23:55 ut , when the density and temperature were enhanced , the magnetic field strength reduced and the magnetic field rotation ceased .
our identification of the mc is the same as that by @xcite .
the mc and the shock sheath ahead of it caused a double - peak major geomagnetic storm with the provisional @xmath7 value of @xmath8 nt ( see the last panel of fig.[fg_wind ] ) .
the mc is the interplanetary counterpart of the cme originating on march 15 because of the following two reasons .
( 1 ) if the mc corresponds to the cme , the transit time of the cme from its first appearance in the lasco / c2 and the arrival at 1 au is about 59.5 hours , and the average transit speed is about 690 km s@xmath4 , which is consistent with the cme speed , 817 km s@xmath4 , in the fov of lasco / c3 and the measured mc speed , @xmath25 km s@xmath4 , at 1 au .
( 2 ) except for the preceding slow cme mentioned in the last section , there was no other cme candidate during march 14 16 responsible for the mc according to the lasco observations .
the slow cme propagated far away from the ecliptic plane and was probably on the other side of the sun .
thus , it should not be detected near the earth .
it should be noted that @xcite also analyzed this event and proposed a different scenario that there were two interplanetary cmes ( icmes ) , corresponding to the slow and fast cmes on march 14 and 15 , respectively , and the main icme , i.e. , the one corresponding to the fast cme , was identified in the interval from about 18 ut on march 17 to 16 ut on the next day .
this interpretation is not in agreement with our above analysis .
the wind data shown in figure [ fg_wind ] reveal that the smooth rotation of magnetic field vector and the bi - directional electron streams ceased at the end of march 17 .
there are various models developed to reconstruct mcs from one - dimensional in - situ data , including cylindrically symmetrical force - free flux rope models ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , asymmetrically cylindrical ( non)force - free flux rope models ( e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and torus - shaped flux rope models ( e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
here we use the velocity - modified cylindrical force - free flux rope ( vfr ) model , which considers the propagation and expansion of a mc as well as the plasma poloidal motion inside the mc @xcite , to fit the mc . in this model , the magnetic field is described by the lundquist solution , and the velocity is incorporated under the assumptions of self - similar evolution and magnetic flux conservation .
this model is proven to yield similar results as the cylindrically symmetrical force - free flux rope model by @xcite .
we choose this model because the fitting results contain some kinematic information of the cme , which may provide some clues on the trajectory of the cme in interplanetary space .
in addition , we also apply the grad - shafranov ( gs ) reconstruction technique @xcite to the mc to see the similarity and difference between the model results . [ cols="^,^,^,^,^,^,^,^,^,^,^,^ " , ] from the left to the right , the columns give ( 1 ) the model used to derive the parameters , ( 2 ) the magnetic field strength at the mc s axis in units of nt , ( 3 ) the radius of the mc in units of au , ( 4 ) the elevation angle of the mc s axis in the gse coordinates , ( 5 ) the azimuthal angle of the mc s axis in the gse coordinates , ( 6 ) the handedness of the mc , ( 7 ) the closest approach of the observational path to the mc s axis , ( 810 ) the propagation velocity of the mc in the gse coordinates in units of km s@xmath4 , ( 11 ) the expansion speed of the mc in units of km s@xmath4 and ( 12 ) the poloidal speed of the mc plasma in units of km s@xmath4 .
the best - fit values of the free parameters of the vfr model for the mc of interest are given in table [ tb_par ] , and the fitting curves are plotted as the dashed blue lines in figure [ fg_wind ] , which match the observed profiles fairly well in both magnetic field and velocity . in particular , we highlight the following parameters : ( i ) sign of the helicity or handedness of the mc is @xmath26 , which obeys the pattern that the southern hemisphere of the sun usually accumulates positive helicity ( e.g. , * ? ? ? * ) .
( ii ) closest approach is @xmath27 @xmath28 , where @xmath28 is the radius of the mc , indicating that the observational path is far away from the mc s axis .
( iii ) orientation of the mc s axis is @xmath29 and @xmath30 in gse coordinates , corresponding to a tilt angle of about @xmath31 when projected on the plane - of - the - sky .
considering that the magnetic polarity in the cme source region is negative / positive on the southern / northern side of the filament associated with the cme ( see fig.[fg_source]c ) and the helicity of the mc is positive , this value is close to the cme s tilt angle of @xmath12 estimated from the gcs fitting .
the orientation also suggests that the angle between the axis of the observed portion of the mc and the sun - earth line is about @xmath22 , meaning that the flank of the mc was passed through ( e.g. , * ? ? ?
( iv ) combination of ( i ) and ( iii ) gives how the magnetic field lines wind in the mc , which is in agreement with the distribution of the magnetic field polarities in the cme source region , i.e. , positive / negative polarity region on the north - west / south - east side of the associated filament ( fig.[fg_source]c ) .
the results of the gs reconstruction are shown in figure [ fg_gs ] and the fitting parameters are listed in table [ tb_par ] too .
it is a substantially different technique from the vfr fitting which is a type of forward models .
the gs reconstruction neither presets the shape of the flux rope nor assumes a fore - free state , but uses the gs equation to infer the two and a half dimensional distribution of magnetic field at the cross - section of the flux rope from the observed magnetic field and thermal pressure along the spacecraft path under the assumption of being time - stationary and magnetohydrostatic .
the fitting of @xmath32 , the sum of the axial magnetic pressure and the thermal pressure , which is essential to the gs reconstruction , yields a residue @xmath33 @xcite as a measure of goodness of fit . as judged from the right panel of figure [ fg_gs ] , the interpretation of a flux - rope configuration shown in the left panel is valid for the limited region within the white contour line and under the assumption that significant axial current exists at the center .
the gs reconstruction gives the same handedness as vfr model , and we can read from figure [ fg_gs ] that the magnetic field at the center is about 23 nt and the radius about 0.08 au in @xmath34-axis or 0.05 au in @xmath35-axis , slightly smaller than but comparable to those derived from the vfr model .
the closest approach is about @xmath36 au .
for the orientation , the gs model is quite consistent with the vfr model in the elevation angle , but deviates significantly in the azimuthal angle .
the angle between the orientations of the gs model and the vfr model is about @xmath37 , the largest inconsistency between the two models .
the comparison suggests that the solution of the fit to the in - situ measurements is not unique .
the most sensitive parameter is the orientation of the flux rope axis . besides
, the selection of the boundaries of a mc might also significantly affect the fitting results ( private communication with k. marubashi and q. hu ) .
it is difficult to evaluate which one is more reliable .
we list the two possible solutions here for reference and also to raise the question for further attention .
@xcite performed ` blind tests ' by applying five different fitting techniques , including the cylindrical linear force - free flux rope model , the elliptical cross - section non - force - free flux rope model and the gs model , to a mhd simulated mc .
the largest deviation among these model results is in the orientation , especially when the observational path is far away from the mc s axis .
the march 17 mc encountered the earth with the closest approach of @xmath27 @xmath28 , falling into this scenario .
thus , it is not surprising that we get quite different results in the orientation from the different models .
however , the tests by @xcite do suggest that the fitting technique based on a cylindrical force - free flux rope is a useful tool .
the in - situ measurements of the solar wind velocity reveal that there were significant components of the velocity in @xmath38 and @xmath39 directions in the gse coordinates .
the vfr model suggests that the @xmath34-component of the propagation velocity of the mc is @xmath40 km s@xmath4 , and the @xmath41-component is about @xmath42 km s@xmath4 .
according to the tilt angle of the cme , the part of the cme closer to the sun - earth line was above the ecliptic plane , and therefore the @xmath34- and @xmath41-component velocities do imply that the cme was approaching the sun - earth line and the ecliptic plane .
since the @xmath41-component velocity is smaller than the @xmath34-component velocity and the longitude is more important than the latitude in this case , we only consider the @xmath34-component velocity in the following analysis .
it was shown by @xcite in the statistical study of 72 mcs that the propagation velocity perpendicular to the radial direction is 19 km s@xmath4 on average , and only 8% of the events had a perpendicular velocity larger than 60 km s@xmath4 .
thus , the @xmath34-component propagation velocity in this case is significant enough to indicate an eastward deflection of the mc from the radial direction . by comparing to the radial propagation speed
, we may infer that the deflection rate in the ecliptic plane , @xmath43 , is about @xmath44 .
assuming that the cme kept the deflection rate all the way from @xmath5 , where the cme left the fov of the lasco / c3 ( fig.[fg_gcs]g ) , to 1 au , the deflection angle can be calculated by the equation 17 in @xcite , i.e. , @xmath45 .
the change of the cme s propagation direction estimated by this method is shown as the blue crosses in figure [ fg_mc_top]b .
such an eastward deflection made the path of the earth cutting through the cme closer to the cme center , and therefore enhanced the geoeffectiveness of the initially west - oriented cme .
it is interesting to think about the cause of the possible eastward deflection of the cme .
first , the fast stream following the cme clearly modified the mc structure by , e.g. , increasing the strength of the magnetic field in the rear portion of the mc , but we think that it probably is not the major cause of the eastward deflection .
this cme is a structure embedded in a co - rotating interaction region ( cir ) characterized by the fast stream behind the cme and the slow stream ahead of it .
one may imagine that the overtaking fast stream would push the cme toward the west . the solar image , e.g. , fig .
[ fg_source]a , also suggests that the cme should be deflected toward the west because of the presence of the large ch on the south - east of it .
however , it is probably not the case as implied by the in - situ data and the vfr model results .
thus , we think that the cme s trajectory was actually controlled by the slow stream ahead of it .
the velocity difference between the cme and the preceding slow stream is about 10 times ( roughly estimated by eyes from the in - situ data ) of that between the cme and the following fast stream .
based on the dips picture @xcite , such velocity differences may lead to a more significant pile - up of solar wind plasma and magnetic field ahead or on the west of the cme than that behind or on the east of it , and therefore cause an eastward deflection .
now we use the dips model to infer the cme trajectory with the constraints of the solar and in - situ observations . to run the dips model , we need the cme propagation speed and the ambient solar wind speed , both of which are the functions of the heliocentric distance . as in @xcite , the ambient solar wind speed is obtained from the 3-dimensional ( 3d ) mhd simulation @xcite for carrington rotation 2161 which covered the period from february 29 to march 27 .
we choose the simulated solar wind speed along the latitude of @xmath46 and within the longitude of @xmath47@xmath22 , along which the cme propagated , to generate the averaged solar wind speed ( the black plus signs in fig.[fg_mc_top]a ) .
the cme propagation speed is derived by the drag - based model ( dbm , * ? ? ?
* ; * ? ? ?
* ) , starting at the distance of @xmath48 with the initial speed of the cme leading edge @xmath49 km s@xmath4 . the simulated solar wind speed mentioned above
is adopted .
we adjust the drag coefficient @xmath50 to match the dbm model output , the arrival time of the mc and the speed , with the in - situ observations at 1 au .
after several trials , the best value of @xmath50 is found to be about @xmath51 km@xmath4 , with which the modeled mc arrival time is 13:32 ut and the speed of the mc leading edge is 622 km s@xmath4 at 1 au , close to the observed arrival time 13:05 ut and measured speed of about 600 km s@xmath4 .
it should be noted that the speed of the cme leading edge consists of two components : the propagation speed and the expansion speed .
the vfr model has suggested that the expansion speed is approximated to be one tenth of the cme radial propagation speed .
the ratio of the expansion speed to the radial propagation speed will hold as long as the cme self - similarly evolved with a constant angular width . by deducting the expansion speed from the speed of the cme leading edge
, we obtain the cme propagation speed as the function of the distance ( see red diamonds in fig.[fg_mc_top]a ) . the red diamonds in figure
[ fg_mc_top]b show the trajectory of the cme derived from the dips model .
it is suggested that , from @xmath5 to 1 au , the cme was deflected by about @xmath6 toward the east , in agreement with the deflection angle , @xmath52 , estimated by the vfr model based on the in - situ observations .
it is difficult to evaluate the error of the predicted deflection angle .
the direct error comes from the uncertainties in both the cme and solar wind speeds . by considering an uncertainty of @xmath53 in them ,
the predicted deflection angle of this cme is about @xmath54 .
however , since dips is a kinematic model , the error may also come from other unknown factors which are not included in the model . in the previous study by @xcite , the deflection angle of the slow cme originating on 2008
september 12 is predicted as @xmath55 , which is much smaller than the expected value , @xmath56 , derived from the observations .
thus the same problem is also applicable to the fast cme investigated here .
here , we study the fast cme originating on 2015 march 15 , which caused the largest geomagnetic storm so far in solar cycle 24 . with the aid of forward modeling ,
the kinematic properties of the cme in the corona are obtained based on the solar imaging data . within 30 @xmath9
the cme propagated west of the sun - earth line along the longitude of about @xmath21 and the latitude of about @xmath57 .
the propagation speed is about 817 km s@xmath4 .
its angular width is wide enough to make its east flank overlap the sun - earth line . on the other hand ,
the in - situ data at 1 au suggests the flank of a mc arriving at the earth .
two different models are applied to the mc to infer the configuration of the mc .
it is found that the handedness is consistent but the orientation of the flux rope is quite different between the models . as suggested by @xcite , such a large deviation in the orientation
is probably due to the spacecraft being too far away from the mc s axis .
currently , it is still difficult to evaluate the reliability of the model results . due to the lack of the stereo observations ,
the propagation of the 2015 march cme in the heliosphere is unclear .
we then try to recover the interplanetary evolution process of the cme from the information at the two ends : near the sun and at 1 au .
the vfr model results based on the in - situ data suggest that the cme experienced a significant eastward deflection with the ratio of @xmath43 of about @xmath44 , implying a @xmath6-deflection toward the earth .
the detailed trajectory of the cme between the two ends are further reconstructed by using the numerical simulation ( for background solar wind ) , dbm model ( for the cme propagation speed ) and dips model ( for the cme trajectory ) under the constraints of the cme s kinematics obtained from the solar and in - situ observations .
the reconstructed trajectory is bent toward the earth , quite consistent with the deflection implied by the vfr model .
this eastward deflection pushed the center of the initially west - oriented cme closer to the earth and probably contributed to the unexpected strong geoeffectiveness of the cme .
however , the lack of interplanetary observations causes that the above inference for this case can not be fully validated though it sounds reasonable , and the origin of this strong geomagnetic storm is still somewhat mysterious .
some models applied in this study can be run and tested online .
one can go to http://space.ustc.edu.cn/dreams/ for the dips model and the vfr model , and to http://oh.geof.unizg.hr/dbm/dbm.php for the dbm model .
we acknowledge the use of the data from soho , sdo and wind spacecraft and the @xmath10 images from kanzelhoehe observatory .
soho is a mission of international cooperation between esa and nasa , and sdo is a mission of nasa s living with a star program .
we also acknowledge the discussion of the event with t. berger , n. gopalswamy , p. hess , y. liu , k. marubashi , c. moestl , t. rollett , m. temmer , c .- c . wu and s. yashiro , under the isest program .
we thank the anonymous referees for their constructive comments .
this work is supported by grants from nsfc ( 41131065 , 41421063 ) , cas ( key research program kzzd - ew-01 and 100-talent program ) , and the fundamental research funds for the central universities .
is also supported by nsfc 41574165 , c.s . by nsfc 41274173 ,
f.s . and z.y .
by nsfc 41474152 and most 973 key project 2012cb825601 , and r.l . by nsfc 41222031 .
b.v . and t.z .
acknowledge the financial support by croatian science foundation under the project 6212 solar and stellar variability " .
feng , x. , s. t. wu , f. wei , and q. fan , a class of tvd type combined numerical scheme for mhd equations with a survey about numerical methods in solar wind simulations , _ space sci .
_ , _ 107 _ , 4353 , 2003 .
gopalswamy , n. , m. shimojo , w. lu , s. yashiro , k. shibasaki , and r. a. howard , prominence eruptions and coronal mass ejection : a statistical study using microwave observations , _ astrophys .
j. _ , _ 586 _ , 562578 , 2003 .
isavnin , a. , a. vourlidas , and e. k. j. kilpua , three - dimensional evolution of flux - rope cmes and its relation to the local orientation of the heliospheric current sheet , _ sol .
_ , _ 289 _ , 21412156 , 2014 .
kilpua , e. k. j. , j. pomoell , a. vourlidas , r. vainio , j. luhmann , y. li , p. schroeder , a. b. galvin , and k. simunac , stereo observations of interplanetary coronal mass ejections and prominence deflection during solar minimum period , _ ann .
_ , _ 27 _ , 44914503 , 2009 .
liu , y. d. , h. hu , r. wang , z. yang , b. zhu , y. a. liu , j. g. luhmann , and j. d. richardson , plasma and magnetic field characteristics of solar coronal mass ejections in relation to geomangetic strom intensity and variability , _ astrophys . j. lett .
_ , _ 809 _ , l34(6pp ) , 2015 .
lugaz , n. , j. n. hernandez - charpak , i. i. roussev , c. j. davis , a. vourlidas , and j. a. davies , determining the azimuthal properties of coronal mass ejections from multi - spacecraft remote - sensing observations with stereo secchi , _ astrophys .
j. _ , _ 715 _ , 493499 , 2010 .
mulligan , t. , and c. t. russell , multispacecraft modeling of the flux rope structure of interplanetary coronal mass ejections : cylindrically symmetric versus nonsymmetric topologies , _
j. geophys .
106(a6 ) _ , 10,58110,596 , 2001 .
shen , f. , x. feng , and w. b. song , an asynchronous and parallel time - marching method : application to the three - dimensional mhd simulation of the solar wind , _ science in china series e : technological sciences _ , _ 52 _ , 28952902 , 2009 .
shen , f. , x. s. feng , y. wang , s. t. wu , w. b. song , j. p. guo , and y. f. zhou , three - dimensional mhd simulation of two coronal mass ejections propagation and interaction using a successive magnetized plasma blobs model , _ j. geophys
_ , _ 116 _ , a09,103 , 2011 .
vrnak , b. , t. ic , t. v. falkenberg , c. mstl , s. vennerstrom , and d. vrbanec , the role of aerodynamic drag in propagation of interplanetary coronal mass ejections , _ astron . &
_ , _ 512 _ , a43 , 2010 .
wang , y. , p.
z. ye , s. wang , g. p. zhou , and j. x. wang , a statistical study on the geoeffectiveness of earth - directed coronal mass ejections from march 1997 to december 2000 , _ j. geophys .
107(a11 ) _ , 1340 , doi:10.1029/2002ja009,244 , 2002 .
wang , y. , c. chen , b. gui , c. shen , p. ye , and s. wang , statistical study of coronal mass ejection source locations : understanding cmes viewed in coronagraphs , _ j. geophys
_ , _ 116 _ , a04,104 , doi:10.1029/2010ja016,101 , 2011 .
wang , y. , z. zhou , c. shen , r. liu , and s. wang , investigating plasma motion of magnetic clouds at 1 au through a velocity - modified cylindrical force - free flux rope model , _ j. geophys .
_ , _ 120 _ , 15431565 , 2015 .
webb , d. f. , n. u. crooker , s. p. plunkett , and o. c. st .
cyr , the solar sources of geoeffective structures , in _ space weather _ , edited by s. paul , j. s. howard , and l. s. george , geophys .
125 , pp . 123142 , agu , 2001 .
wood , b. e. , j. l. lean , s. e. mcdonald , and y .- m .
wang , comparative ionospheric impacts and solar origins of nine strong geomagnetic storms in 2010 - 2015 , _ j. geophys .
_ , _ accepted _ , doi:10.1002/2015ja021,953 , 2016 .
yermolaev , y. i. , m. yermolaev , g. zastenker , l. zelenyi , a. petrukovich , and j .- a .
sauvaud , statistical studies of geomagnetic storm dependencies on solar and interplanetary events : a review , _ planet .
space sci .
_ , _ 53 _ , 189196 , 2005 . zuccarello , f. p. , a. bemporad , c. jacobs , m. mierla , s. poedts , and f. zuccarello , the role of streamers in the deflection of coronal mass ejections : comparison between stereo three - dimensional reconstructions and numerical simulations , _ astrophys .
, _ 744 _ , 66(14pp ) , 2012 . | the largest geomagnetic storm so far in the solar cycle 24 was produced by a fast coronal mass ejection ( cme ) originating on 2015 march 15 .
it was an initially west - oriented cme and expected to only cause a weak geomagnetic disturbance .
why did this cme finally cause such a large geomagnetic storm ?
we try to find some clues by investigating its propagation from the sun to 1 au .
first , we reconstruct the cme s kinematic properties in the corona from the soho and sdo imaging data with the aid of the graduated cylindrical shell ( gcs ) model .
it is suggested that the cme propagated to the west @xmath0@xmath1@xmath2@xmath3 away from the sun - earth line with a speed of about 817 km s@xmath4 before leaving the field of view of the soho / lasco c3 camera .
a magnetic cloud ( mc ) corresponding to this cme was measured in - situ by the wind spacecraft two days after the cme left lasco s field of view . by applying two mc reconstruction methods
, we infer the configuration of the mc as well as some kinematic information , which implies that the cme possibly experienced an eastward deflection on its way to 1 au .
however , due to the lack of observations from the stereo spacecraft , the cme s kinematic evolution in interplanetary space is not clear . in order to fill this gap
, we utilize numerical mhd simulation , drag - based cme propagation model ( dbm ) and the model for cme deflection in interplanetary space ( dips ) to recover the propagation process , especially the trajectory , of the cme from @xmath5 to 1 au under the constraints of the derived cme s kinematics near the sun and at 1 au .
it is suggested that the trajectory of the cme was deflected toward the earth by about @xmath6 , consistent with the implication from the mc reconstruction at 1 au .
this eastward deflection probably contributed to the cme s unexpected geoeffectiveness by pushing the center of the initially west - oriented cme closer to the earth . |
there are many examples , both man made and naturally occurring , of phenomena which begin with a small disturbance and end in a catastrophe . such events include electrical network failures @xcite , forest fires @xcite , avalanches @xcite , nuclear chain reactions @xcite , snapping ropes @xcite and landslides @xcite .
a common feature of the distribution of the sizes of such events , and the mathematical property which links the very small to the very big , is power - law scaling .
such scaling often appears over a number of orders of magnitude with the end of the power law region marked by an exponentially decaying probability density , referred to as an `` exponential cut off '' @xcite .
the values of the power law exponent , and the cut
off point will depend on the nature of the system , but often not on its fine details a phenomenon known as universality @xcite .
the physical origin of the cut
off may be the physical size of the system or its inherent ability propagate the cascade . for infinite systems
, there exists a critical level of instability at which the distribution is a pure power - law . as the system approaches this critical point , the cut off moves increasingly rapidly toward infinity , and the expected cascade size diverges . in this work it is our aim to investigate the phenomenon of power law crossover in cascade size distributions .
crossover occurs when the distribution of large cascades follows a different power law to that of small cascades .
this type of behaviour was discovered @xcite in the distribution of bursts of snapping events in bundles of fibres under tension and close to complete breakdown .
it is our aim to show that similar behaviour may be exhibited by a system in which cascades occur repeatedly over an extended period .
each cascade has the effect of increasing the stability of the system so that subsequent cascades propagate less freely .
such a stabilizing effect is seen , for example , in regions prone to forest fires where the extent and frequency of large fires has been reduced by planned burning of limited areas @xcite .
another example is where the spread of disease through a population is reduced by the presence of recovered ( or vaccinated ) individuals , who act as firewalls preventing transmission @xcite .
we might also expect to see a similar effect in terrain susceptible to landslides .
in fact , there is some evidence of crossover behaviour in landslide size distributions @xcite .
however , our model is not tied to one particular physical system , but rather it puts forward a generic explanation of how crossover behaviour might arise .
our approach is to produce the simplest possible model ( and explanation ) of driven cascades , that is mathematically tractable , and that preserves some key features of real cascading phenomena
. the fundamental quantity of interest to us will be the ability of the system to propagate a cascade , which will be described by a single number , @xmath0 , the propagation power .
in physical terms , this quantity might be determined by the connectivity of a network of unstable nodes , the density of unstable material in a system , or the average proximity of the components of a system to failure .
because we view the propagation power as a measure of how easily cascades propagate between parts of system it will not depend on the absolute volume of the system it is an intensive property .
the dynamical properties of @xmath0 will be determined by two processes .
first , @xmath0 will tend to increase over time , but with an element of unpredictability , described by brownian noise @xcite so that in the absence of cascades : @xmath1 where @xmath2 is the mean rate of increase of @xmath0 , @xmath3 is a standard wiener process @xcite , and @xmath4 controls the magnitude of the noise .
the physical origin of such a process in , for example , a forest fire model might be the drying out of vegetation due to unpredictable weather , or in a landslide model , the natural variability of pore water pressure .
we assume that the magnitude of the noise is independent of system size , and therefore it represents an external driving process .
the second influence on @xmath0 arises from the cascades themselves , which act to stabilize the system . to capture this stabilizing effect
, we suppose that if the size of the @xmath5th cascade since an arbitrary reference time is given by the continuous random variable @xmath6 , then in response @xmath0 changes by @xmath7 .
the parameter @xmath8 measures the sensitivity of @xmath0 to cascades , and will depend on the size of the system , and the stabilizing effect on those parts of it that are involved in the cascade . for convenience , we will refer to @xmath8 as the `` inverse system size '' .
note that the distribution of @xmath6 will itself depend on @xmath0 .
we define @xmath9 to be the sum of all cascade sizes that have occurred since @xmath10 and let @xmath11 count the number that have occurred .
we then have that : @xmath12 together with the driving dynamics ( [ driving ] ) we have a complete stochastic differential equation for @xmath13 : @xmath14 we will assume that cascades begin when a small part of the system spontaneously fails
. since a larger system will have more potentially unstable material then the rate at which cascades begin should scale in proportion to the system size @xmath15 .
assuming that the cascade rate is constant and independent of the cascade history , then the times between cascades will be exponentially distributed and @xmath11 will be a poisson process with intensity @xmath16 , where @xmath17 is a constant of proportionality , independent of system size . without loss of generality in the model we can set @xmath18 by re scaling time and adjusting @xmath19 and @xmath20 .
as we mentioned above , the distribution from which individual cascades sizes , @xmath6 , are drawn will depend on the value of @xmath0 when they occur .
we write the probability density function of this distribution @xmath21 . in section [ casc ]
we introduce a continuous state branching process to describe cascades , which we cap at size @xmath22 ( this cap is proportional to system size , and ensures that the propagation power can not become negative ) .
we find that : @xmath23 where @xmath24 as @xmath0 approaches a `` critical value '' @xmath25 . at this critical point , @xmath21 is a pure power
law and the mean cascade size diverges . for @xmath26 , @xmath27
so the power
law region is `` cut off '' at approximately @xmath28 . when @xmath29 , the mean cascade size increases further still and is infinite in the limit @xmath30 .
the divergence of the mean cascade size as the critical point is approached from below will create a self stabilizing effect which pushes the system away from @xmath25 .
the combination of this automatic stabilization and the upward drift of the driving process means that the propagation power will fluctuate about a mean value lying in the subcritical region .
we will find that the magnitude of fluctuations is controlled largely by @xmath20 for large systems . of central interest to us
is the long term cumulative record of cascade sizes in the system , which we describe with a probability density function @xmath31 .
this will be an average of @xmath21 over the values of @xmath0 at which cascades take place .
we define @xmath32 to be the probability density function for the value of the propagation power at time @xmath33 .
the `` steady state '' density function for @xmath0 is then @xmath34 .
we then have that @xmath35 this expression may also be thought of as the probability density function for the size of the next cascade observed after some arbitrary ( but large ) time .
one of our main results is to show analytically that fluctuations in @xmath0 about its typical subcritical value , causing it to approach temporarily closer to criticality , are what generates crossover behaviour in the `` averaged '' cascade distribution @xmath31 . from the above arguments it is clear that typical value of @xmath0 , corresponding to the peak of @xmath36 , will be increased by increasing the driving rate @xmath19 . in the limit of infinite system size
we will find that the stabilizing effect of diverging mean cascade size near @xmath25 means that it is always the case that @xmath37 .
therefore , in the limit @xmath38 the system will sit at the critical point .
similar behaviour is exhibited by the forest fire model @xcite , where the equivalent of @xmath19 is the rate of tree growth . because there are a wide range of parameter values for which both our model and the forest fire model are near critical ,
both may be seen as examples of `` self organised criticality '' ( s.o.c . )
the requirement that @xmath19 be sufficiently large in order for the system to be near criticality means that our model does not exhibit s.o.c . in its purest form , as seen in the sandpile @xcite
. however , we can still draw analogy between the steady increase in @xmath0 and the addition of energy or particles in truly self organising models .
suppose that each cascade begins when a small volume of the system experiences a `` failure '' event ( for example it may catch fire , explode , get infected or begin motion ) .
this may induce further parts to fail and so on , forming a sequence of failures whose volumes @xmath39 are referred to as _
generations_. we assume that the generations follow a _ continuous state branching process _ @xcite .
the relationship between the sizes of the successive generations is encoded by an _ offspring _ distribution , @xmath40 .
this is the probability distribution for the amount of failed material that each unit of the current generation triggers in the next .
we assume that @xmath40 remains the same throughout the cascade .
suppose that the zeroth generation of the process , @xmath41 , is an integer , then the size , @xmath42 , of the first generation is the sum of @xmath41 independent copies of a @xmath40-distributed random variable . in this case
the distribution of @xmath42 is simply the convolution of @xmath40 with itself @xmath41 times , written @xmath43 . if @xmath41 is not an integer , we extend the idea of @xmath44-fold convolution , following seneta and vere jones @xcite , as follows .
suppose that @xmath45 is a random variable drawn from the distribution @xmath40 .
the function : @xmath46 is the laplace transform of @xmath40 . given this definition ,
@xmath47 is the laplace transform of @xmath48 where @xmath49 . by relaxing the constraint that @xmath44 be an integer
, we may define the laplace transform of @xmath42 , conditional on @xmath41 , to be @xmath50 , and then extend this rule to later generations : @xmath51 this recursive equation defines the relationships between the distributions of successive generations , giving a complete characterisation of the process once @xmath40 is defined .
we will take @xmath40 to be the gamma distribution @xcite @xmath52 which has density function @xmath53 defined for @xmath54 , where the variables @xmath55 and @xmath56 are referred to as the shape and scale parameters .
our choice of @xmath40 is motivated by the requirement that it have a physically plausible shape ( not bimodal ) , have defined moments , lead to mathematical tractability , and that @xmath57 $ ] for all @xmath58 .
in fact , the asymptotic behaviour of the distribution of the total size of the cascade : @xmath59 will depend only on the mean and variance of the offspring distribution so we expect our crossover predictions would hold for other forms of @xmath40
. a significant advantage of our particular choice is that for all real @xmath60 , @xmath61 , from which we see , via equation ( [ lapbranch ] ) , that our gamma branching process is defined by the relationship @xmath62 where @xmath41 is the volume of the first generation of the cascade . in equation ( [ bp ] )
the symbol @xmath63 means `` is distributed as '' . using this recursive definition of the cascade , it is possible to compute the distribution of the total cascade size @xmath64
however , we first fix the relationship between the offspring distribution and @xmath0 .
we will allow this dependence to enter through the mean , @xmath65 , of the offspring distribution , so that the average volume contributed to each successive generation of the cascade per unit volume of the previous generation is proportional to @xmath0 .
we choose to let @xmath66 and leave @xmath5 as a free parameter which controls the variance and shape of @xmath40 .
we will write the probability density function for @xmath64 as @xmath21 and consider first the case where there is no limit on how large cascades can become . we may calculate @xmath21 by interpreting @xmath64 as the first passage time of a random walk .
however , the details of this calculation are somewhat technical , and are not part of our main story , so we confine them to appendix [ csbp ] .
the resulting probability density function for @xmath64 is : @xmath67 \frac { e^{- \kappa z } } { z^{3/2 } } \text { as } z \rightarrow \infty \label{psiasymp}\end{aligned}\ ] ] where : @xmath68 in the context of equation ( [ psiasymp ] ) the symbol @xmath63 means `` tends asymptotically to '' . to the author
s knowledge , equation ( [ psi ] ) is a new result in the theory of continuous state branching processes @xcite .
considering equation ( [ psiasymp ] ) , when the mean of the offspring distribution is one ( @xmath69 ) , then @xmath70 , so the distribution is asymptotically a pure @xmath71 power law with infinite moments .
the value of propagation power @xmath72 at which this behaviour occurs is referred to as the critical point .
when @xmath73 the quantity @xmath74 gives the location of the exponential cut - off .
provided @xmath75 , the distribution @xmath21 is normalised and its moments are defined .
it is useful to have explicit expressions for the first two moments of @xmath21 in this case . in appendix [ csbp ]
we show that , provided the cascade size is not limited by finite system size , then @xmath76 when @xmath77 the distribution ( [ psi ] ) is not normalized .
this is referred to as the `` supercritical '' regime where infinitely large cascades become possible .
the total probability weight is equal to @xmath78 . in appendix [ csbp ]
we show that : @xmath79 where : @xmath80}(p)\ ] ] and @xmath81 is one of the two real branches of the lambert w function @xcite , the other being @xmath82 .
we will cap cascades at a maximum size @xmath83 so that they can not cause the propagation power to take negative values .
this cap scales in proportion with the system size . assuming that cascades will stop abruptly once @xmath84 , then the tail of @xmath21 will replaced with a delta function @xmath85 where @xmath86 is the total probability weight in the tail , plus the probability of an infinite cascade : @xmath87
the capped cascade distribution is therefore : @xmath88}(z ) \psi_p(z ) + \omega(z_{max } ) \delta(z - z_{max}),\ ] ] and the @xmath44th cascade moment in a finite system is : @xmath89 where the second term arises from integration over the delta function in the capped cascade density . for simplicity , for the remainder of the paper we will begin all cascades with a failure of volume @xmath90 .
for arbitrary @xmath8 , the steady state distribution of the propagation power equation ( [ sde ] ) can not be found analytically , but may be determined by simulation .
however , for small @xmath8 , jumps which are not small with respect to the system size become increasingly rare , and the dynamics of @xmath0 may be approximated by diffusion with drift , giving rise to a pure diffusion equation which is analytically tractable . in the limit @xmath30 , this approximation becomes exact . to illustrate the rarity of large jumps , figure [ sim ] shows a simulation of @xmath13 when @xmath91 , @xmath92 , @xmath93 and @xmath94 . in this case out of @xmath95 cascades only six exceeded @xmath96 of the system size , and the system reached criticality only once out of all recorded times .
simulation of @xmath13 using parameter values are @xmath91 , @xmath92 and @xmath93 and @xmath97 giving a critical @xmath0 value of @xmath98 .
values of @xmath13 were recorded when the jump which preceded them exceeded @xmath99 of the system size .
, width=7 ] to derive our diffusion approximation we view @xmath0 as evolving in discrete steps , with its value being recorded immediately before each cascade .
each change in @xmath0 is therefore comprised of a cascade , followed by a period of diffusion until the next cascade occurs .
this defines a discrete time stochastic process @xmath100 where @xmath101 is the propagation power immediately before the the @xmath44th cascade .
we let @xmath102 be the probability density function for @xmath101 .
we then define @xmath103 to be the probability density function for the size , @xmath104 , of the next single step of the process given that the current value of propagation power is @xmath105 .
note that @xmath103 is independent of @xmath44 .
the master equation governing the evolution of @xmath102 is then : @xmath106 our aim is to find the steady state distribution : @xmath107 .
before we continue our analysis of the master equation we will make clear the link between the distributions of discrete time process , and the underlying continuous time process ( [ sde ] ) . in section [ intro ] , we defined @xmath36 to be the steady state probability density function for the continuous process , @xmath13 , which may intuitively be thought of as the density function for the value of @xmath0 observed at an arbitrary time @xmath108 , large enough so that the influence of initial conditions is insignificant . due to the properties of exponential waiting times @xcite , the time interval , @xmath109 , between @xmath108 and the last cascade before @xmath108 will be exponentially distributed . letting @xmath110
be the value of @xmath0 immediately after this last cascade , we see that @xmath111 .
now let @xmath112 be another large time , but restricted to the set of cascade times so that we are making an observation of the discrete time process . again using the properties of exponential waiting times , the time @xmath113 since the previous cascade will have the same distribution as @xmath109 (
intuitively this arises because @xmath108 is likely to lie in a larger than average waiting interval ) . because the value of @xmath0 just after the previous cascade will be drawn from the same distribution as @xmath114 , we have that @xmath115 where @xmath116 denotes equality in distribution . as a result ,
observations of the discrete time and the continuous time process have the same distribution at large times : @xmath117 if the times between cascades were not exponentially distributed then the probability distribution of the state of the system immediately preceding a cascade would not , in general , be the same as the distribution at a randomly selected time . in that case equation ( [ meancasc ] ) would be incorrect because @xmath21 must be averaged over the distribution of @xmath0 immediately preceding a cascade . to approximate the steady state solution to ( [ master ] ) we derive the corresponding kramers moyal equation @xcite by expanding the integrand of ( [ master ] ) in powers of @xmath104 : @xmath118 substituting this approximation back into the master equation we find that : @xmath119 where we have made use of the normalisation of the step size distribution to simplify the first term .
we now define the first two moments of step size to be : @xmath120 as @xmath30 , both the time between cascades and their effect on @xmath0 decline so the step distribution becomes increasingly sharply peaked about @xmath121 . we therefore ignore moments of higher order than two and obtain a discrete time analogue of the fokker
planck equation for pure diffusion : @xmath122 we are interested in the steady state behaviour of the process , where @xmath123 , and in particular we require the function @xmath36 , which from equation ( [ fp ] ) , satisfies : @xmath124 = \frac{1}{2 } \partial_{pp}\left[b(p)f(p)\right]\ ] ] the first integral of this equation , the probability current @xmath125 , will be zero . in the limit of large system size , for @xmath73 ,
the probability of cascades which engulf the entire system is zero , and the system can never access the region @xmath126 because the mean cascade size diverges in the neighborhood of @xmath127 .
we therefore expect our diffusion approximation to become exact . as this limit
is approached our expressions for @xmath128 and @xmath129 take the asymptotic form : @xmath130 \\ b(p ) & \sim \epsilon \sigma^2 + \epsilon^2 \left[\mathbf{e}(z^2 ) - 2\mu \mathbf{e}(z ) + 2 \mu^2\right ] \label{basymp}\end{aligned}\ ] ] in equation ( [ basymp ] ) we have retained the @xmath131 term because the moments of the cascade distribution become very large as @xmath132 . for any finite system
we expect our approximation to break down near this critical point , and to breakdown globally if there is sufficient probability weight in the supercritical region where infinite cascades have nonzero probability .
we will explore this breakdown using simulations .
the first moment of the cascade distribution has asymptotic behaviour : @xmath133 which diverges near @xmath134 , creating an infinite negative drift .
the probability weight at the critical point therefore declines to zero as @xmath30 and the divergence in @xmath129 will never be realised .
we may therefore drop the @xmath131 term for infinite systems . making use of @xmath128 and the simplified @xmath129
we see from equation ( [ j ] ) that the limiting form of @xmath36 satisfies : @xmath135 although the cascade distribution @xmath21 is not defined for @xmath136 , in the interests of tractability , we will take equation ( [ feqn ] ) as valid over the interval @xmath137 $ ] , yielding the following expression for @xmath36 : @xmath138 } ( 1-k p)^{\frac{2}{k \sigma^2 } } \exp\left[-\frac{2 \mu}{k \sigma^2}(1-kp)\right]\ ] ] for all parameter values of interest to us , the probability weight in the invalid region @xmath139 $ ] is less than @xmath140 .
note that this solution is independent of @xmath8 , but we expect it to become an increasingly good approximation to the true solution as @xmath30 for values of @xmath73 .
by making use of the large system size approximation to the propagation power probability density , @xmath36 , together with the asymptotic form of the cascade distribution ( [ psiasymp ] ) , is it possible to investigate the @xmath141 asymptotic behaviour of the mean cascade distribution in the limit @xmath30 . throughout this section
all asymptotic formulae hold as @xmath142 . to simplify our expressions
we define a new parameter : @xmath143 the mean cascade distribution may then be written @xmath144 where @xmath145 we have extended the lower limit of integration to @xmath146 for tractability , since the integrand will be negligible when @xmath136 .
the integral ( [ psibar ] ) is not tractable .
however , as @xmath142 , the weight of the integrand becomes concentrated in a shrinking neighborhood of the critical point .
we may therefore approximate the integral asymptotically by replacing the first exponent in the integrand with its taylor expansion to quadratic order about @xmath72 : @xmath147 making the change of variables @xmath148 , our approximation becomes : @xmath149 we now make a second change of variables : @xmath150 which gives : @xmath151 dt.\end{gathered}\ ] ] we may extract the asymptotic behaviour of this integral by noting that , for finite @xmath33 , as @xmath142 the @xmath152 dependent exponential term tends to one .
we now note that provided @xmath153 : @xmath154.\ ] ] the asymptotic behaviour of the mean cascade distribution is therefore a pure power law : @xmath155 where @xmath156\mathcal{c}.\ ] ] so , in the limit of large system size , the tail of the mean cascade distribution , rather than being exponential , follows a power law with an exponent which is an increasing function of the variance of the destabilisation process . for smaller @xmath152 ,
when most of the probability weight in @xmath36 corresponds to cut - offs at larger @xmath152 values , we see @xmath157 behaviour preserved .
together with our asymptotic predictions this gives rise to a power law crossover .
the theoretical distribution @xmath31 , in the limit @xmath30 , is illustrated in figures [ mu ] and [ sig ] together with the asymptotic result ( [ asymp ] ) .
log log plot of the mean cascade distributions ( dashed lines ) @xmath31 in the limit @xmath30 when @xmath158 , @xmath93 and @xmath19 takes the values @xmath159 and @xmath160 .
the larger @xmath19 value produces a crossover point at a higher value of @xmath152 .
also shown are the asymptotic power law predictions ( [ asymp ] ) ( solid lines ) and the function @xmath157 as a dotted line . , width=8 ] log log plot of the mean cascade distributions ( dashed lines ) @xmath31 in the limit @xmath30 when @xmath158 , @xmath161 and @xmath20 takes the values @xmath162 and @xmath163 .
the larger @xmath20 value produces a shallower tail in the cascade distribution .
also shown are the asymptotic power law predictions ( [ asymp ] ) ( solid lines ) and the function @xmath157 as a dotted line . , width=8 ]
using simulations we will now test the validity of our large system crossover predictions , explore the influence of @xmath8 and the effect of the model parameters on the averaged cascade distribution , @xmath31 .
we make use of two simulation methods ; a `` naive '' technique where every cascade is simulated , and an `` accelerated '' technique which uses a diffusion approximation when the probability of jumps of significant size in comparison to @xmath15 is sufficiently small . the simplest method to determine the averaged cascade distribution is to simulate the stochastic process described by equation ( [ master ] ) .
the results ( shown in figure [ naive ] ) were obtained by simulating @xmath164 cascades , and recording their sizes in bins of increasing width .
it is clear from the figure that to fully examine the tail behaviour of the distribution would require us to simulate significantly more cascades .
this would be prohibitively time consuming so we adopt an accelerated scheme for the remainder of our results .
the circles show the simulated cascade distribution @xmath31 when @xmath165 , @xmath92 , @xmath166 and @xmath167 .
the results were obtained by simulating the propagation power process over @xmath164 cascades .
the black line shows the theoretical cascade distribution in the limit @xmath30 .
, width=8 ] for the parameters used in figure [ naive ] we see that our theoretical cascade distribution in the limit @xmath30 was accurate for cascades up to at least @xmath168th of the system size . due to the exponential cut - off in @xmath21 , large cascades are only likely when @xmath0 is in the vicinity of the critical point .
the behaviour of the distribution of @xmath0 in this region determines the tail behaviour of the average cascade distribution .
given that the diffusion approximation works well for smaller @xmath0 values , we may speed up our simulation by explicitly simulating cascades only once we are close to @xmath25 . for lower @xmath0 values
we approximate the process as pure diffusion , which is much faster to simulate .
we determine the switch point @xmath169 by requiring that for @xmath170 , the probability of observing at least one cascade that is larger than @xmath168th the system size over the simulation must be effectively zero ( taken to be @xmath140 ) . for @xmath170
we advance @xmath13 using a fixed time step @xmath171 .
for @xmath172 we simulate every cascade , where the times between them are exponentially distributed with mean @xmath8 .
typically , @xmath171 may be taken to be orders of magnitude larger than @xmath8 , but must be small enough so that the process does nt significantly `` overshoot '' @xmath169 into the region where the quality of the diffusion approximation is uncertain .
we refer to the sum total of long ( @xmath171 ) and short ( mean length @xmath8 ) time steps to be the number of steps in the simulation .
provided @xmath173 , then , because the system spends most of its time in the region @xmath170 , we can effectively simulate @xmath13 over a much larger time interval than with the naive method .
rather then generating a catalogue of cascades , this technique produces an approximate density function for @xmath0 .
we average the cascade distribution over this approximate density function in order to find @xmath174 . in figure [ accel ]
we have used this accelerated method to investigate the cascade distribution for a larger system than in figure [ naive ] but with otherwise identical parameter values . also plotted
is our asymptotic prediction ( [ asymp ] ) , and the cascade density @xmath21 when @xmath0 is equal to its mean value .
this serves to highlight the difference between crossover behaviour and exponential cut off .
we note that the analytic predictions are well matched in this case ; the diffusion approximation remained valid some distance beyond the switch point @xmath169 .
the circles show the simulated cascade distribution @xmath31 when @xmath175 , @xmath92 , @xmath166 and @xmath167 .
the results were obtained by simulating the propagation power process using the accelerated technique over @xmath164 time steps .
the black line shows the theoretical cascade distribution in the limit @xmath30 .
the dashed lines are pure power laws .
the shallow gradient line has exponent @xmath71 , whereas the steep gradient line is our asymptotic prediction ( [ asymp ] ) .
the dotted line shows @xmath21 when @xmath0 is equal to its mean value .
note the presence of the exponential cut
off.,width=8 ] we now investigate the breakdown of our crossover predications by simulating smaller systems . in figure [ breakdown ] we have simulated two different sized systems that are both smaller than in figure [ accel ] .
we note that when @xmath176 the distribution of the first generation of the cascade possesses a maximum located away from zero and the cascade distribution inherits this characteristic . for the smaller system
, the crossover fails to fully develop , and the exponent begins to increase again .
this occurs because the distribution of propagation power does not decay to zero at the critical point , so the averaged cascade distribution includes significant contributions from values of @xmath0 for which @xmath21 is approximately a pure power law for @xmath177 .
the breakdown of the diffusion approximation occurs in part because the frequency of cascades is insufficient to realise their divergent mean size on short time scales . in the larger system where @xmath178 we see that the crossover develops more fully .
because cascades occur with greater frequency , fluctuations in the short term average of the cascade size are reduced .
the open circles show the simulated cascade distribution @xmath31 when @xmath91 , @xmath179 , @xmath180 and @xmath181 .
the filled circles show the simulated cascade distribution when @xmath182 and all other parameters are the same .
the results were obtained by simulating the propagation power process using the accelerated technique over @xmath164 time steps .
the solid line shows the theoretical cascade distribution in the limit @xmath30 .
, width=8 ] in figures [ mu ] and [ sig ] , we illustrate the role of the parameters @xmath19 and @xmath20 in determining the behaviour of the cascade distribution @xmath31 in the limit of large system size . figure [ mu ] shows that for fixed @xmath20 , the driving rate @xmath19 , which determines the location of the maximum of @xmath36 , fixes the location of the crossover . at larger driving rates , the crossover point shifts to larger cascade sizes .
this is because , for larger @xmath19 , the peak of @xmath36 is nearer to the critical point .
therefore the typical exponential cut off scale is larger and the @xmath71 scaling region is extended . in figure [ sig ] we see that a noisier driving process reduces the magnitude of the tail exponent , but the size of the @xmath71 region remains unaffected .
increased variability in the driving process means that although the typical distance from the critical point is unchanged , the system spends more time in close proximity to it and therefore large cut
offs are more heavily weighted . in both figures
the asymptotic power
law predictions ( [ asymp ] ) for the cascade distribution give an indication of the size of the region over which the crossover occurs .
we have presented a driven cascade model which , in the limit of large system size , exhibits power law crossover behaviour in its cascade size distribution . for smaller systems , the crossover partially develops , but the distribution moves back toward the initial @xmath71 power law at larger cascade sizes , because the system is able to reach and exceed the critical value of propagation power . the mechanism which generates the crossover is a competition between the driving process , which increases the instability of the system , and the cascade process , which reduces it .
as the propagation power nears the critical point , @xmath25 , the mean cascade size diverges , so that in the limit of large system size , @xmath25 is not accessible for finite @xmath19 .
the upper tail of the distribution of propagation power , @xmath36 , therefore decays to zero at @xmath25 .
the asymptotic behaviour of the averaged cascade distribution , @xmath31 , is determined by averaging the asymptotic behaviour of the cascade distribution for fixed propagation power , @xmath21 , over the upper tail of @xmath36 .
the result is a power law with exponent lower than @xmath71 , producing a crossover .
we suggest that the presence of a crossover in a cascade size distribution may indicate that the system is able to self stabilize through frequent but non catastrophic cascades , and that the mechanism which drives the instability of the system to cascading failure is inherently noisy . the stabilizing cascades act to prevent the system from reaching a fully critical state where very large cascades can occur . because our system will be near critical for a wide range of parameter values ( large enough @xmath19 ) , it may be considered to exhibit `` self organised criticality '' . however , the crossover is more obvious if the typical value of propagation power is not too close to criticality , and it has been our focus to explore crossover .
it should also be noted that an infinite system will only become fully critical in the limit @xmath38 , whereas a finite system may become critical or supercritical through fluctuations .
the location of the crossover indicates how close to criticality the system will typically be found , because it is determined by the peak of @xmath36 .
it remains to adapt the ideas contained in our simple model to investigate real physical systems such as forest fires and landslides .
the purpose of this appendix is to derive the probability density function for the total cascade size in the continuous state branching process defined by relationship ( [ bp ] ) , and also to determine the probability of an infinite size cascade .
we begin by showing that the our problem may be interpreted as a first passage time problem .
consider the branching process @xmath39 defined by the relationship @xmath183 with @xmath41 given . in order to calculate the distribution of @xmath184 ,
we show that @xmath64 may be viewed as the first passage time of a stochastic process through the origin .
we first give the definition of the `` gamma process '' @xcite , which takes place in continuous time on @xmath185 $ ] .
if @xmath186 is a gamma process with parameters @xmath5 and @xmath187 then : we now define a new stochastic process @xmath192 , where @xmath41 is the size of the first generation of our branching process . by defining @xmath193 to be the cumulative cascade size up to the @xmath44th step
, we may show that the processes @xmath194 and @xmath195 have the following relationship : @xmath196 provided that the cascade has not ended for some @xmath197 . here
@xmath116 denotes equality in distribution .
we may deduce that this relationship holds inductively .
we note first that from the defining properties of the gamma process : @xmath198 assuming that relationship ( [ map ] ) holds for all @xmath199 then @xmath200 since @xmath201 then ( [ map ] ) holds when @xmath202 and therefore for all @xmath44 by induction .
the cascade ends at the first generation for which @xmath203 , at which point @xmath204 , so the total size of the cascade is equal to the first time that the process @xmath194 meets the origin . in order to find the cascade size distribution we need to solve this first passage time problem .
our first passage time problem is most easily solved by viewing the stochastic process @xmath194 as the limit of a discrete state random walk . here
we show how the appropriate random walk is constructed .
we begin by noting that the negative binomial distribution , which has probability mass function : @xmath205 provides an arbitrarily close discrete approximation to the gamma distribution for appropriate choice of the parameters @xmath104 and @xmath206 .
the approximation is set up in the following way .
we divide @xmath185 $ ] into a discrete lattice of constant spacing @xmath207 , and let @xmath208 be a discrete random variable which approximates @xmath209 . let @xmath208 have the probability mass function : @xmath210 the mean and variance of @xmath211 are then @xmath212 and @xmath213 .
letting @xmath214 and @xmath215 be the values of @xmath104 and @xmath206 for which the mean and variance of @xmath216 and @xmath211 are equal , we find that : @xmath217 with these choices of @xmath104 and @xmath206 , in the limit @xmath218 the discrete distribution converges to @xmath52 in the following sense : @xmath219 = \frac{x^{k-1}e^{-\frac{x}{\theta}}}{\gamma(k ) \theta^k},\ ] ] where the notation @xmath220 represents the integer part of the real number @xmath221 . as well as
providing an arbitrarily good approximation to the gamma distribution , the negative binomial distribution also has the property that if @xmath222 is a negative binomial variable , then the sum of @xmath223 independent copies of @xmath45 has distribution @xmath224 .
this allows us to set up a discrete approximation to the gamma process as follows .
we divide each unit of time into @xmath225 subunits and let the random variable @xmath226 have distribution : @xmath227 we will refer to this as the `` atomic '' random variable with `` atomic distribution '' . we may approximate the gamma process as a sum of a sequence , @xmath228 of independent copies of @xmath226 : @xmath229 in other words , we are viewing @xmath186 as a scaled random walk who s step sizes are given by the atomic distribution .
we may approximate the process @xmath194 in a similar way to @xmath186 : @xmath230 let @xmath231 and @xmath232 be the first passage time of the walk ( [ qwalk ] ) , starting from position @xmath223 . considering the first step , which will have size @xmath233 , we have that : @xmath234 since the time for the walk to get from position @xmath223 to the origin is equal to the time it takes to get to position @xmath235 plus the time to get from position @xmath235 to the origin , then @xmath236 . the quantity @xmath232 is therefore the sum of @xmath223 independent copies of @xmath237 .
we have from ( [ rec ] ) that : @xmath238 if @xmath239 and @xmath240 are the probability generating functions for @xmath237 and @xmath226 , then from equation ( [ zrec ] ) we have @xmath241 \\ & = s \mathbf{e } [ ( h(s))^a ] \\ & = s f(h(s)).\end{aligned}\ ] ] from the negative binomial mass function we have that : @xmath242 we are interested in the probability generating function of @xmath232 , which is just @xmath243 .
the coefficient of @xmath244 in this function may be determined using the lagrange inversion formula @xcite : @xmath245 h^m(s ) & = \frac{1}{n}[h^{n-1}]\left\ { \left(\frac{d}{dh } h^{m}\right ) f^n(h)\right\ } \\ & = \frac{m}{n } [ h^{n - m } ] f^n(h ) \\ & = \frac{m}{n } \frac{\gamma(n(1+\delta r^*)-m)}{\gamma(n\delta r^ * ) \gamma(n - m+1)}(1-q^*)^{\delta r^*n } ( q^*)^{n - m } \\ & = \mathbf{p}\{z_\delta(m ) = n\}\end{aligned}\ ] ] where the notation @xmath246f(x)$ ] stands for the coefficient of @xmath247 in the taylor series of @xmath248 .
we now have the probability mass function for the cascade size in the discrete branching process which approximates the continuum process that we are interested in .
now that we have the solution to the discrete problem we solve the continuous problem by taking the continuum limit .
we find expressions for the moments of the cascade size using a similar method .
using a martingale method we determine the probability that the cascade is of finite size in the supercritical case .
we obtain the continuum cascade density function , which we will call @xmath249 , by setting @xmath250 and @xmath251 and then taking the limit @xmath218 of @xmath252 : @xmath253 the asymptotic properties of @xmath249 may be determined by making use of stirling s approximation : @xmath254 .
the result is : @xmath255 \frac { e^{- \kappa z } } { z^{3/2 } } \text { as } z \rightarrow \infty\ ] ] where @xmath256 provided @xmath257 , the distribution @xmath249 is normalised and its moments are defined .
it is useful to have explicit expressions for the first two moments of @xmath249 in this case .
we may compute the moments of the ( discrete ) distribution of @xmath237 by differentiating the generating function relationship : @xmath258 , and then solving for @xmath259 and @xmath260 . using the expression for @xmath240 , together with the fact that when @xmath261 , @xmath262 , we find that : @xmath263 the total cascade size , @xmath64 , has the same distribution as the sum of @xmath264 copies of @xmath237 , so @xmath265 and @xmath266 , yielding the first two moments of the cascade distribution in exact form : @xmath267 numerical integration of the exact distribution ( [ exact ] ) reveals that it is not normalized when @xmath268 .
this is the `` supercritical '' regime . in general the total probability weight is equal to @xmath78 , which is less than one in the supercritical case because there is a non zero probability of seeing an infinite cascade .
we may deduce this probability by considering the stochastic process : @xmath269 taking the expectation value of this process , conditional on its value at @xmath10 we find that : @xmath270 if we let @xmath206 be the solution to the equation @xmath271 then this expectation will be independent of time .
the value of @xmath206 which solves this equation is : @xmath272 letting @xmath64 be the first time at which the process meets the origin , then we have that @xmath273 this gives the result presented in equation ( [ mom ] ) .
the author would like to acknowledge murad banaji , samia burridge , alexey kuznetsov , andreas kyprianou and malcolm whitworth for useful discussions , as well as the anonymous referees for careful reading and constructive criticism . | we propose a model which explains how power law crossover behaviour can arise in a system which is capable of experiencing cascading failure . in our model
the susceptibility of the system to cascades is described by a single number , the _ propagation power _ , which measures the ease with which cascades propagate .
physically , such a number could represent the density of unstable material in a system , its internal connectivity , or the mean susceptibility of its component parts to failure .
we assume that the propagation power follows an upward drifting brownian motion between cascades , and drops discontinuously each time a cascade occurs .
cascades are described by a continuous state branching process with distributional properties determined by the value of the propagation power when they occur . in common with many cascading models ,
pure power law behaviour is exhibited at a critical level of propagation power , and the mean cascade size diverges .
this divergence constrains large systems to the subcritical region .
we show that as a result , crossover behaviour appears in the cascade distribution when an average is performed over the distribution of propagation power .
we are able to analytically determine the exponents before and after the crossover . |
caustics of light have been known since ancient times .
the rainbow is a common example of a light caustic that forms when the family of refracted light rays is projected on to the plane of the sky .
another common example is the heart shaped , or nephroid pattern seen on the bottom of a polished coffee cup .
thus caustics of light are regions where the light intensity is very large .
caustics have well defined physical properties .
the light intensity on the inner ( concave ) side of the rainbow varies like the inverse of the square root of distance from the rainbow .
however , when the rainbow is approached from the outside ( convex ) side , the light intensity remains small until the rainbow is reached , at which point it shoots up abruptly .
the coffee cup caustic is shown in fig .
1 . the caustic is the envelope of the family of reflected light rays , i.e. the curve tangent to all members of the family of rays .
there are two qualitatively distinct regions separated by the caustic : one with three rays at each point and the other with one ray at each point .
consider the point @xmath0 .
near this point , the light intensity varies @xmath1 when measured along @xmath2 , and @xmath3 when measured along @xmath4 .
the intensity at the point @xmath5 is double the intensity at the point @xmath6 for small @xmath7 .
near the smooth curve and far from the point @xmath8 , the light intensity only varies in the region with 3 rays at each point , like the inverse square root of the distance from the smooth curve . as long as one is far from the boundary between the two regions , a small change in position results in only a small change in light intensity .
however , close to the boundary , a small change in position leads to a very large change in light intensity .
what we have described are called _
catastrophes_. caustics are made up of sections of catastrophes ( more precisely , we are concerned with the bifurcation set of the catastrophe ) .
the rainbow caustic is a fold catastrophe .
the coffee cup caustic of fig .
1 consists of 2 fold catastrophes that join together forming a cusp catastrophe .
their properties are described by catastrophe theory which is a branch of singularity theory applied to physical phenomena .
light forms caustics because light is nearly collisionless .
dark matter also has this property .
we may thus expect dark matter in galactic halos to form caustics .
dark matter caustics are singularities in physical space in the limit of zero velocity dispersion @xcite . in reality
, the finite velocity dispersion of the dark matter particles cuts off the divergence in the density .
dark matter caustics are then regions of high density .
we may expect a continuous infall of dark matter to form caustics provided ( i ) the particles are collisionless and ( ii ) they have negligible velocity dispersion . the shells or arcs seen around some giant elliptical galaxies @xcite are caustics in the distribution of stars . when a small galaxy is disrupted , the stars of the small galaxy fall into the gravitational potential of the giant galaxy
these stars are then sub - virial , i.e. they are cold . being compact , stars are nearly collisionless .
the infall of cold collisionless stars leads to the formation of caustics at the turnaround radii .
the observations of caustics of stars strongly suggests the existence of caustics of dark matter due to the infall of cold , collisionless particles . these are called _
outer caustics _ and appear as a series of thin spherical shells surrounding galaxies . besides the outer caustics , the infall of dark matter particles results in _
inner caustics_. when the particles possess angular momentum , they do not pass through the center , and hence have non - zero inner turnaround radii .
inner caustics form near these locations .
2 shows the dark matter trajectories for the case of ( a ) radial infall and ( b ) infall with angular momentum .
when the trajectories are exactly radial , the infall forms a point singularity at the center .
when a small amount of angular momentum is added to the flow , the result is a caustic made up of sections of catastrophes .
further small perturbations do not change the qualitative nature of the inner caustic in ( b ) .
the inner caustics are made up of folds and cusps which may terminate on higher order catastrophes .
the formation of inner caustics is described in detail in @xcite .
qualitatively , the caustic of fig .
2(b ) resembles the coffee cup caustic of fig .
1 . the spherical shell caustics mentioned earlier qualitatively resemble the rainbow caustic .
one may ask whether substructure in real galactic halos can greatly increase the number of flows and weaken the caustics . in @xcite , it was shown that each substructure clump can break up a flow into 3 sub - flows , called `` daughter flows '' .
downstream of @xmath9 clumps , the number of sub - flows @xmath10 can be very large for moderately large @xmath9 @xcite .
however this does not destroy the discreteness of the phase space hypersurface because the density of the main flow is not equally divided among the sub - flows .
let us consider the very first flow , i.e. dark matter particles falling towards the halo center for the first time .
the first turnaround marks the radius at which particles bound to the halo decouple from the hubble expansion .
a caustic does not form at the first turnaround radius since there is only one flow there .
the first outer caustic occurs at the _ second _ outer turnaround radius .
the first inner caustic occurs at the _ first _ inner turnaround radius .
we thus see that the first inner caustic forms before the first outer caustic . as an example , let us assume that the first flow contributes @xmath11 to the local dark matter density in the absence of gravitational scattering by substructure .
let us assume there is enough substructure to break up the first flow into a large number , say @xmath12 of sub - flows .
let us also ( incorrectly ) assume that the density is equally divided among the sub - flows .
we then have @xmath12 flows , each contributing @xmath13 to the local dark matter density , and each producing an inner caustic of no realistic importance near the inner turnaround radii .
the flows eventually fall out of the potential well ( i.e. their radial velocities becomes positive ) , and form @xmath12 outer caustics , of negligible density contrast at the outer turnaround radii .
the conclusion then would be that there are _ no _ physically relevant caustics , inner or outer .
this conclusion contradicts the observations of @xcite who have obtained evidence for the existence of a dark matter outer caustic in their study of galaxies in the group around ngc 5846 , and have interpreted the caustic as that of second turnaround , i.e. the first outer caustic .
this result has been interpreted by @xcite as evidence that an inner caustic exists in that group .
indeed it is hard to understand how the first outer caustic can form if the infalling flow is so diffused that it can not produce the first inner caustic . in @xcite
, it was found that gravitational scattering of dark matter particles by inhomogeneities does not greatly affect the first few flows unless most of the dark matter is composed of very massive ( @xmath14 ) clumps .
we know of no evidence that the dark matter in the milky way is composed of such massive clumps . on the contrary ,
the recent findings of @xcite show that @xmath15 of the dark matter in halos is accounted for by smooth accretion of particles not bound in halos . assuming a continuous infall of cold dark matter particles with angular momentum ( assumed aligned with the baryon angular momentum ) , one may expect to find a series of inner caustics in the plane of the galaxy .
authors @xcite ( motivated by the work of @xcite ) have proposed a model of the milky way halo that predicts the locations of the caustics , and the approximate density near the caustics . in this model
, the earth naturally lies between two inner caustics .
the distance to the nearest inner caustic has been estimated in @xcite .
a ring like feature in the cluster cl 0024 + 1654 was studied by @xcite who have interpreted the feature as a dark matter inner caustic .
the presence of inner caustics in spiral galaxies was investigated by @xcite , who found possible evidence for their existence by analyzing the rotation curves of many galaxies .
rises in the rotation curve of the milky way also show possible evidence for the existence of caustics in our galaxy @xcite . in @xcite , the presence of a triangular feature in the infrared spectrum
was proposed as possible evidence for the presence of a nearby inner caustic .
certain high resolution simulations see cold streams @xcite that are associated with inner caustics
. however , recent n - body simulations performed by @xcite do not find significant inner caustics .
the absence of inner caustics in these simulations is inconsistent with the predictions of @xcite and with the claimed observational evidence . in this article , we do not attempt to reconcile the simulations with the observations .
instead we suggest that a nearby dark matter caustic may be revealed by the velocity distribution of dark matter particles in the solar neighborhood .
if there exists a nearby dark matter inner caustic , the dark matter velocity distribution is influenced by the cold flow forming the caustic .
we show that the annual modulation expected in the recoil rate measured by a detector can be used to detect the presence of a nearby inner caustic , even for weak density enhancements .
it was shown in @xcite that the continuous infall of cold collisionless matter necessarily produces an inner caustic .
let us review the argument : consider a sphere with a conveniently chosen radius , such that all particles of a given flow pass through the sphere .
then , we may identify every particle of the flow by 3 parameters @xmath16 .
@xmath17 is the time when the particle crossed the sphere on its way in to the halo . @xmath18 and @xmath19 are co - ordinates on the sphere .
let @xmath20 be a location in physical space where the dark matter density is measured .
a catastrophe forms at points where the mapping from the space @xmath16 to the space @xmath21 is singular .
the condition for the mapping to be singular at @xmath21 is the vanishing of the jacobian determinant : d = det = ( ) = 0 .
[ caustic_condition ] let @xmath22 be the point on the reference sphere where the angular momentum is a maximum ( the angular momentum field defined on a sphere has a maximum somewhere unless it is zero everywhere ) .
let us choose @xmath23 to be the point of closest approach of the particles with the most angular momentum , and let us define @xmath24 .
then we have at @xmath25 : = x = 0 = x = 0 = x = 0 .
the first condition is due to the reversal of the sign of the radial velocity at @xmath25 .
the other two conditions are due to the angular momentum maximum at @xmath22 .
since the four vectors @xmath26 , @xmath27 , @xmath28 , and @xmath29 can not all be mutually perpendicular at the same point , at least two of the vectors are linearly dependent , or at least one of them is zero .
this satisfies the caustic condition eq .
[ caustic_condition ] .
thus the infall of a cold , collisionless shell of particles necessarily results in the formation of an inner caustic , regardless of assumptions of symmetry . the structure of the inner caustic depends on the spatial distribution of angular momentum on the initial reference sphere . for purely curl like initial conditions ,
the inner caustic resembles a ring , made up of sections of the elliptic umbilic catastrophe @xcite . for purely gradient like initial conditions , the inner caustic has a more complicated shape and is made up of sections of the hyperbolic umbilic catastrophe @xcite .
for the special case of axial symmetry with gradient like initial conditions , one may find swallowtail and butterfly catastrophes @xcite .
dark matter caustics can affect the distribution of stars in the galaxy . in @xcite
, it was proposed that the monoceros ring of stars may have formed due to the the presence of the @xmath30 inner caustic of the milky way .
two mechanisms were found by which the star density could be enhanced in the vicinity of an inner caustic , and these may have played a role in the formation of the monoceros ring .
authors @xcite examined the possibility of observing caustics by gravitational lensing , while @xcite have studied dark matter annihilation in caustics .
these possibilities however require high densities in the vicinity of the caustic .
let us now review the annual modulation expected in the recoil rate of a detector , and show that it is an excellent probe of a nearby dark matter caustic even when the density enhancement due to caustic is low .
it was shown by @xcite that due to the motion of the earth about the sun , the flux of dark matter particles reaching the earth is modulated with a period of 1 year , and is largest at certain times of the year .
dama is a dark matter direct detection experiment that measures the annual modulation in the recoil rate , and claims a positive result with @xmath31 confidence @xcite .
unfortunately , it seems the results obtained by dama are incompatible with the null results of other experiments ( see for example @xcite ) .
the idea of using the annual modulation as a probe of a nearby dark matter caustic is not new .
authors @xcite have calculated the expected recoil rate for the late infall self similar caustic model . the effect of cold streams has been studied by several authors ( see for example @xcite ) . here
we review the known results , and study the case of a maxwellian velocity distribution modified by a single cold flow due to the presence of a nearby caustic .
we present results for two cases ( i ) when the stream contributes significantly to the dark matter density ( i.e. caustic with a large density boost ) and ( ii ) when the stream is a small perturbation .
it is shown that even in case ( ii ) , the effects are potentially observable .
let us consider weakly interacting massive particles ( wimps ) as the dark matter .
the number of recoils per unit time per unit nuclear mass is given by dr = dq , where @xmath32 is the mass of the nucleus of the detector , @xmath33 is the atomic mass number , @xmath34 is the nucleon mass , @xmath35 is the dark matter density at the location of the detector , @xmath36 is the scattering cross section , @xmath37 is the recoil energy , and @xmath38 is the velocity of the dark matter particles with respect to the detector . for elastic scattering ,
the energy transferred to the nucleus is given by q = ( 1 - ) , [ q ] where @xmath39 is the wimp - nuleus reduced mass and @xmath18 is the scattering angle in the center - of - momentum frame . from eq .
[ q ] , we see that in order to observe a nuclear recoil with energy @xmath37 , the wimps must posses a _ minimum _
speed v_min = , [ vmin ] since @xmath40 .
it is possible for wimps to have a larger speed than the minimum given in eq .
[ vmin ] , but they can not have a smaller speed for a given recoil energy @xmath37 . here , we consider only the spin - independent cross section .
the differential cross section @xmath41 is expressed in the form : = f^2(q ) = f^2(q ) .
@xmath42 is the nuclear form factor .
it is obtained from nuclear physics experiments @xcite , and contains the momentum dependence of the cross section : f(q ) = e^- ( qs)^2 , in units where @xmath43 and @xmath44 are set to 1 .
@xmath45 , @xmath46 fm , @xmath47 fm , @xmath48 , and @xmath49 is the spherical bessel function .
it is easy to see that @xmath50 for small @xmath37 , and falls off at large @xmath37 .
we therefore have the number of recoils per unit energy , per unit time , per unit detector mass : = _ _ v_min(q)^dv .
let us define the two quantities : t(q , t ) & = & _ v_min(q)^dv m(q_1,q_2,t ) & = & _ q_1^q_2 dq f^2(q ) t(q , t ) .
@xmath51 is called the mean inverse speed and @xmath52 is the number of recoils per unit time , per unit detector mass , between energies @xmath53 and @xmath54 .
let us now focus our attention on the distribution of wimp velocities .
the spectrum of dark mater velocities is discrete , i.e. a sum over flows : f(v ) = _ i ( v - v_i ) , [ flows ] where @xmath55 is the velocity of the @xmath56 flow .
there are a large number of flows that occupy the inner region of phase space , while the fast flows which are few and well separated in velocity , are mostly found in the outer regions of phase space
. a detector with finite resolution will likely not be able to resolve the numerous flows in inner phase space .
it is useful then to break up the sum over all flows into 2 parts : a sum over cold flows , and a sum over thermal flows .
we may replace the sum over thermal flows by a maxwell - boltzmann velocity distribution : f_max(v_wh ) = , [ max ] where we have ignored the effect of the finite escape velocity .
the subscript @xmath57 stands for `` wimp - halo '' and indicates that the velocities are measured in the rest frame of the halo , not the detector .
we take @xmath58 to be 220 km / s for the milky way .
. [ flows ] can be expressed as : f(v ) = _ cold i ( v - v_i ) + f_max(v ) = f_cold(v ) + f_max(v ) .
let us now compute the mean inverse speed @xmath51 for the two distributions . consider a single cold flow : @xmath59 , where the subscript @xmath60 stands for `` flow - halo '' and indicates that the flow velocity @xmath61 is measured relative to the halo .
transforming to the earth s rest frame using @xmath62 , we obtain @xmath63)$ ] .
@xmath64 is the flow velocity relative to the sun , and @xmath65 is the velocity of the earth relative to the sun .
the speed of the flow relative to the earth is given by |v_fe(t)| = |v_f - v_(t)| & = & ^1/2 & & v_f , [ flow_earth ] for @xmath66 .
the mean inverse speed is : t_flow(q , t ) & = & & = & .
[ t_flow ] @xmath67 is the unit step function = 1 for @xmath68 and 0 otherwise .
@xmath69 is the time averaged value of the earth s velocity about the sun .
let us now consider the maxwellian distribution given by eq .
we again transform to the rest frame of the detector using the relation @xmath70 to obtain the speed ( 1-dimensional ) distribution @xmath71 : f(v ) = , giving a mean inverse speed t_max(q , t ) = , [ tmax ] with @xmath72 given by v_eh(t ) & = & ^1/2 & & v_. as we mentioned in the introduction , the continuous infall of dark matter results in a series of outer and inner caustics .
if the angular momentum of the dark matter is aligned with that of the baryons , these caustics can be expected to lie in the galactic plane . as a result
, the earth lies between two inner caustics . in the self - similar infall model of @xcite ,
the earth is situated close to the fifth inner caustic .
the velocity distribution at the earth s location is therefore influenced by the particles forming the caustic .
let us choose a co - ordinate system in which the @xmath73 axis points towards the galactic center , the @xmath74 axis points in the direction of galactic rotation , and the @xmath75 axis points towards the north galactic pole .
@xcite have listed the velocities of the different dark matter flows ( note that the system of co - ordinates used here is different from that used in @xcite ) . here
, we will consider only the two flows producing the caustic closest to the sun , which have velocities relative to the halo @xcite @xmath76 km / s .
relative to the sun , these flows have velocities : v_a & = & 253.6 ( 0.3549 x + 0.9345 y - 0.0276 z ) v_b & = & 261.4 ( -0.4208 x + 0.9067 y - 0.0268 z ) .
[ flow_vel ] the velocities of the sun ( about the halo center ) and the earth ( about the sun ) in these co - ordinates are respectively ( see for example @xcite , and references therein ) : v_&= & 233.3 v_(t ) & = & 29.8 , [ earth ] where the angle @xmath77 .
we note that earth s velocity about the sun is most closely aligned with the sun s velocity about the halo center when @xmath78 , which occurs around june 1 .
the two velocity vectors are most misaligned when @xmath79 which occurs six months later , around nov 30 .
[ fig3 ] shows the two caustic flow velocities @xmath80 , and @xmath81 relative to the earth . using eq .
[ flow_earth ] , [ flow_vel ] , and [ earth ] , we see that @xmath80 is largest when @xmath82 ( around nov 4 ) and smallest when @xmath83 ( around may 5 ) .
@xmath81 is largest when @xmath84 ( around jan 21 ) and smallest when @xmath85 ( around july 22 ) .
one of these flows is the dominant flow , and for definiteness , let us assume flow @xmath86 is the dominant flow .
this flow shows an @xmath87 speed modulation with a maximum of @xmath88 km / s on @xmath89 november 4 , and a minimum of @xmath90 km / s on @xmath89 may 5
. let @xmath91 be the fraction contributed by the dominant caustic flow to the total dark matter density at the earth s location .
then , f(v ) = .
@xmath91 can be large if the density enhancement due to the caustic is large .
@xcite have estimated that the dominant caustic flow , dubbed the `` big flow '' contributes @xmath92 of the dark matter density at the earth s location .
we now show that even if the dominant flow contributes as little as 5% to the local dark matter density , it has potentially observable consequences .
[ fig4 ] shows the mean inverse speed @xmath51 for an assumed wimp mass @xmath93 gev , and a germanium detector with atomic mass number @xmath94 , as a function of recoil energy @xmath37 .
the top panel ( a ) shows @xmath51 for the two extreme cases @xmath95 ( density entirely due to the caustic flow , shown by the two solid curves black and red ) and @xmath96 ( caustic flow absent , shown by the dashed curves blue and pink ) , expected in november , and in june respectively . for a flow velocity relative to the earth @xmath97 , the maximum energy at which recoils are measured is given by q_max = 25 , for the assumed values of @xmath98 , @xmath99 , and @xmath97 , with the maximum and minimum values occurring in november and may respectively .
the height of the step is given by eq .
[ t_flow ] and peaks when the speed is the lowest i.e. in may , and is smallest in november when the speed is greatest .
note that the modulation of the edge of the step ( @xmath100 square of the flow speed ) is twice the modulation of the step height ( inverse of the flow speed ) , and has opposite phase .
the flow ceases to be visible to the detector during part of the year starting from @xmath101 kev ( for the assumed values of @xmath98 and @xmath99 ) . as the energy is increased
, @xmath102 is non - zero close to november , resulting in a maximum then .
let us look at the behavior of the recoil rate integrated over all energies @xmath103 : m(0,,t ) & = & _ 0^dq f^2(q ) t(q , t ) = _ 0^q_max dq & ~ & ~v_f , where we ignored the energy dependence of @xmath104 .
thus , when integrated over all recoil energies , the recoil rate peaks in november for this particular flow , and is smallest in june .
however , when @xmath105 is measured over only small recoil energies , we have : m(0,q_small , t ) = _ 0^q_small dq f^2(q ) t(q , t ) ~ , which implies that the integrated recoil rate peaks in june for small recoil energies @xmath37 . as @xmath37 is increased , the phase abruptly reverses , peaking in november .
the abrupt shift in @xmath102 is because the flow is no longer visible to the detector around june at these energies , owing to the smaller speed around june . in contrast , for a maxwellian velocity distribution , the recoil rate peaks in november for small @xmath37 , reverses phase as @xmath37 is increased , and peaks in june for sufficiently large @xmath37 @xcite .
the two dashed curves show the modulation of @xmath102 for the purely maxwellian case .
the difference is far less dramatic than for the case of the pure caustic flow . at low recoil energies , @xmath102 peaks in november (
blue dashed curve ) and is smallest in june ( pink dot - dashed curve ) .
the phase reverses at @xmath89 15 kev ( for the assumed values of @xmath106 gev and @xmath94 ) , and @xmath102 then peaks in june and has a minimum in november . at high energies , @xmath102 is small since ( i ) there are few particles with sufficient energy to effect recoils and ( ii ) @xmath104 is small for large @xmath37 .
[ fig4](b ) shows the more physical case with both the caustic flow and the thermal halo contributing to the recoil energy spectrum .
shown are the two cases @xmath107 ( strong density enhancement ) and @xmath108 ( weak density enhancement ) .
let us first consider the case @xmath107 ( shown by the solid red and black curves ) . the caustic flow dominates until the flow ceases to be visible to the detector during parts of the year , resulting in a sharp drop in @xmath102 . for the case of @xmath108 ,
the difference in @xmath102 between june and november measurements is not so large . at low energies
, @xmath102 is mostly due to the maxwellian , and is largest in november . with increase in recoil energy , the modulation amplitude decreases , reversing phase at @xmath109 kev .
the peak now occurs in june .
the next change occurs at @xmath110 kev , with @xmath102 measured in november being larger than in june , owing to the absence of the caustic flow near june .
the phase changes again at @xmath111 kev , due to the total absence of the caustic flow . for @xmath112 kev ,
the modulation is consistent with a maxwellian halo , with a maximum measured in june .
[ fig5 ] shows the annual modulation in @xmath113 as a function of the time of year , for the case @xmath107 ( caustic flow contributes 50% to the local dark matter density ) .
the six panels show different recoil energy ranges , with a bin size of 5 kev .
panels ( a ) , ( b ) , and ( c ) are qualitatively identical , and the caustic flow is dominant , with the recoil rate being largest in june , and smallest in november . the caustic flow is not seen during part of the year in panel ( d ) , resulting in the steep minimum around june .
the modulation is about 25% . in panel ( e )
, much of the flow is absent around june , leading to a near @xmath114 modulation in recoil energy .
the flow is completely invisible in panel ( f ) , and the modulation of @xmath115 is due to the maxwellian distribution .
we see from panel ( a ) that @xmath105 peaks in june and is lowest in november , in the @xmath116 kev energy band . if one were to incorrectly assume a maxwellian distribution with no cold flow contribution , panel ( a ) would imply an upper limit for the wimp mass of about @xmath117 gev ( for the assumed @xmath94 ) , far below the value of @xmath93 gev used in this example .
thus , the phase of the annual modulation constrains the wimp mass @xcite only when the distribution of particle velocities in the solar neighborhood is known .
one may however use the results of two different experiments to constrain the mass without assuming a form for the velocity distribution @xcite .
[ fig6 ] shows @xmath113 for the case @xmath108 ( a caustic flow contribution of 5% ) . in the @xmath116 kev energy range
, @xmath105 shows a minimum in june , and a maximum in november due to the dominant contribution of the maxwellian halo .
phase reversal occurs in the @xmath118 kev region , and the modulation is smallest at these energies . in panel ( c ) , @xmath105 peaks in june and is minimum in november .
the caustic flow influences @xmath105 in panels ( d ) and ( e ) because at these energies , the flow is visible to the detector only during certain months of the year . as a result ,
even a @xmath119 contribution from the caustic flow can lead to a @xmath119 modulation effect , comparable to the modulation produced by the dominant maxwellian component . in panel
( e ) , the peak of @xmath105 occurs in november , and the modulation is far from sinusoidal . in panel
( f ) , the caustic flow does not contribute , and the recoil rate is purely due to the maxwellian distribution .
we have discussed the formation of dark matter caustics and provided possible observational evidence for the existence of outer and inner caustics . the self - similar infall model predicts the locations of the inner caustics , and their approximate densities . according to this model and the observations , the earth may be located near the fifth inner caustic .
one may hope to detect the nearby caustic by means of the annual modulation expected in the recoil rate of a dark matter detector .
we discussed the annual modulation effect and derived expressions for the expected recoil rate .
we considered the case of a maxwellian velocity distribution modified by a cold flow due to the nearby caustic in the self - similar infall model .
3 shows the velocities of the two flows that contribute to the formation of the nearest caustic .
4 shows the mean inverse speed @xmath51 expected for a cold flow , for a maxwellian distribution , and for a combination of the two .
we then computed the recoil rate integrated over energies in 5 kev bins .
we considered two cases : ( i ) a 50% contribution due to the caustic flow ( fig .
5 ) and ( ii ) a 5% contribution ( fig 6 ) .
we showed that the phase of the annual modulation at low energies can be used to constrain the wimp mass only if the velocity distribution of dark matter particles in the solar neighborhood is known .
we also showed that even a small contribution by a caustic flow can significantly alter the modulation of the recoil rate , at energies near the threshold energy of the caustic flow .
annual modulation is thus an excellent tool to detect a nearby dark matter caustic . | we discuss the formation of dark matter caustics , and their possible detection by future dark matter experiments .
the annual modulation expected in the recoil rate measured by a dark matter detector is discussed .
we consider the example of dark matter particles with a maxwell - boltzmann velocity distribution modified by a cold stream due to a nearby caustic .
it is shown that the effect of the caustic flow is potentially detectable , even when the density enhancement due to the caustic is small .
this makes the annual modulation effect an excellent probe of inner caustics .
we also show that the phase of the annual modulation at low recoil energies does not constrain the particle mass unless the velocity distribution of particles in the solar neighborhood is known . |
type ii supernovae ( sne ) are classified by the presence of h in their spectra and sub - classes are based on their observed light curves .
they are most likely caused from the gravitational collapse of a main sequence stars with m @xmath2 8 - 10 @xmath3 .
the differences in type ii light curves was first pointed out by pskovskii ( 1978 ) by introducing the beta parameter which resulted in a continuous classification based on the slope of the light curve .
another study showed that light curves can be differentiated into two distinctive sub - classes , plateaus and linears @xcite .
pata et al . (
1994 ) , using a multivariable factor analysis , introduced a new classification based on an absolute peak magnitude vs. @xmath4 graph resulting in three sub - classes ; bright ( includes both plateaus and linears ) , normal ( includes both plateaus and linears ) , and faint .
good observations of plateau sne such as sn 1969l , sn 1987a , sn 1993j and recently sn 1999em and sn 1999gi are usually followed by extensive theoretical light curve ( lc ) studies .
after sn 1969l lc calculations correctly predicted features found in the observations @xcite .
later more refined analytic treatments and hydrodynamical codes were developed @xcite and applied to red supergiant progenitors .
@xcite conducted a numerical parameter study to examine how the ejected mass , progenitor radius , and explosion energy affect the plateau duration and magnitude of the light curve and photospheric velocity of material .
they showed that relations between explosion energy , envelope mass , and progenitor radius can be obtained from observations of plateau duration , the absolute magnitude and the photospheric velocity .
we are currently preparing a paper investigating a comparison between supernovae of evolved numerical stars and polytrope stars using the method @xcite .
more recently hamuy ( 2003 ) has conducted an observational parameter study based on properties of 24 type ii supernova spectra and light curves .
the author finds correlations between ni mass and plateau luminosity , large ranges in all parameters , and a continuum in the parameter space for type ii plateaus .
other studies of type ii plateaus have found similar results .
one study presents a technique of determining ni mass based on h@xmath5 luminosity at nebular phases @xcite .
as mentioned above we will be publishing a paper comparing the light curves of numerical model explosions to those presented in @xcite and compare the least - squares fit formulas in their paper to those found in this study @xcite .
many papers on sn 1987a have been published on lc modeling and references can be found in review articles @xcite . in order to obtain a good fit to sn 1987a variations in envelope masses , explosion energies , progenitor radii , and mixing of both hydrogen - helium and @xmath1ni were explored but only in a limited parameter space @xcite .
podsiadlowski et al .
( 1992 ) in an attempt to find variations in type ii progenitors examined binary systems and showed it is possible to find many different scenarios .
hsu et al . ( 1993 ) , using hydrodynamical models , examined type ii light curves by varying the envelope mass and explosion energy .
the unusual lc of 1993j produced interest in possibly new progenitors .
many studies of sn 1993j conclude that the progenitor was a 4 @xmath3 helium core with an envelope of about 0.2 @xmath3 of hydrogen , a radius of about 300 r@xmath6 and belonged to a binary system @xcite
. a more extensive study on sn 1993j was conducted by blinnikov et al .
( 1998 ) comparing calculations from two different numerical codes by analyzing the evolution and explosion of the progenitor .
young et al.(1995 ) , using the cepheid distance to m81 ( freedman et al . , 1994 ) , found a lower limit to the ni mass of 0.1 . using the x - ray light curve suzuki and nomoto ( 1995 ) found a upper limit to explosion energy to be 1@xmath7 ergs .
recently the massive companion star for sn 1993j was observed 10 years after the explosion , confirming the suggested binary system with two similar mass stars @xcite .
superior observations of supernova progenitors are now starting to constraining model progenitors and making light curve analysis more precise , e.g. sn 2001du @xcite .
two well observed sne 1999em and 1999gi have been extensively studied .
the nature of the progenitor of sn1999em is constrained to be around 12 @xcite .
the upper mass limit of sn 1999gi has been revised to 15 .
both these progenitor masses are similar to that found in numerical fits to the light curves of sn 1999em and sn 1999gi @xcite . as observations become more complete type ii plateaus have a potential to be used as distance indicators .
the distances to both sne were found by using the expanding photosphere method ( epm ) @xcite .
other studies examined type iis as possible distance indicators using expansion velocities that correlate with the bolometric luminosity @xcite .
young and branch ( 1989 ) compared observed type iip light curves on an absolute magnitude scale and found a large spread in absolute magnitude , their plateau duration , and slope of the tails , attributing it to differences in the physical properties of the progenitor .
these studies indicate that the parameter space is quite large for the progenitor of type ii sne and it is difficult to say what type of progenitor will explode or not .
what is needed is a parameter study that encompasses all variables and all combinations , but this is unfeasible .
therefore the aim of this and future studies is to have a standard sn model with parameters that are consistent with knowledge of progenitor stars , explosion mechanisms , and hydrodynamics .
it is then of interest to see if all observed light curves can fit into this model and how far in parameter space they deviate from the standard model . in this study
the standard model is a 20 main sequence star that has gone through a wind mass loss of 2 , left a neutron star of 1.4 , thus ejecting about 16 with a progenitor radius of 3@xmath8 cm .
the energetics of the simulation has a total energy of 1@xmath7 ergs or 1 foe ( ten to the fifty one ergs ) and ejects 0.07 @xmath1ni mixed throughout the 6 he core .
the standard model is similar to sn 1987a except for the larger radius and less extensive mixing . in this paper
the parameter space around this model is explored .
+ in this paper i show the results of how varying each parameter can influence the shape and absolute magnitude of type ii light curves out to 400 days .
the five parameters explored in this study are ; the progenitor radius , envelope mass , explosion energy , @xmath1ni mass , and @xmath1ni mixing . all light curves calculated in this paper use the numerically evolved 20 @xmath0 main sequence model with a 6 helium core from woosley and weaver ( 1980 ) .
the envelope parameters , mass and radius , are varied using homology transformations ( section 2 ) .
the models are then exploded in a one dimensional , flux - limited hydrodynamical code with a simple prescription for gamma - ray deposition ( section 3 ) .
the bolometric light curves are calculated and plotted on an absolute magnitude scale ( section 4 ) .
the results are discussed in section 5 .
for the initial models of all explosions i use the 6 @xmath0 helium core from woosley and weaver ( 1980 ) , originally a 20 @xmath0 main sequence star .
the envelope mass and radius are subsequently modified in a systematic way .
a total of 8 models were constructed that are identified by a specific mass and radius ( table i ) . in this study three
different envelope masses were used producing total masses of 8 , 12 , and 16 @xmath0 . for each envelope mass , three different radii where used 43 r@xmath6 ( @xmath9 cm ) , 430 r@xmath6 ( @xmath10 cm ) , and 4300 r@xmath6 ( @xmath11 cm ) .
the original h rich envelope was modified by homologous transformations to give the various masses and radii @xcite . for a homologous transformation in radius : @xmath12 where r is the old radius , r@xmath13 is the new radius , and @xmath14 is the percent changed . for a homologous transformation in mass : @xmath15 where @xmath16 and @xmath17 are the ratio of the gas pressure to the total pressure for the star before and after the homologous transformation . @xmath18 and @xmath19 are initial and final mass , and @xmath20 and @xmath21 are the mean molecular weights for the initial and final configuration . following either of these calculations the remaining physical variables , @xmath22 and t , must be solved to ensure hydrostatic equilibrium .
the homologous transformation is performed on the hydrogen envelope only and then matched to the he core .
the core and various envelopes were then re - zoned to ensure a continuous density , temperature and radius .
eight progenitor models in all were constructed for this study each having 170 zones in order to have consistent results for opacity floors ( opacity minimum ) and gamma - ray deposition which are used to calculate zone dependent quantities like the photosphere and the gamma - ray contribution to the luminosity ( see section 3 ) .
the innermost 1.4 @xmath3 is assumed to form a neutron star and is removed from the core , but set as an inner boundary condition in the explosions . +
the constructed models are in hydrostatic equilibrium but not necessarily in radiative equilibrium .
this should not be a problem since the shock moves through the star in 6 to 240 hours .
radiative equilibrium is important concerning the validity of the model stars representing real stars , but the point of this study is to compare various simple models and see how the lc responds .
+ three different final explosion energies , @xmath23 , @xmath24 , and @xmath25 ergs are obtained by artificially placing the required energy , divided between kinetic and thermal , in the first few mass zones .
this procedure is justifiable since the explosion mechanism for type ii sne is beyond the scope of this paper .
furthermore , the type of shock is only important in the nucleosynthesis ( aufderheide 1991 ) which is not included in our study .
the radiation - hydrodyanmics code given to the author @xcite contains the following attributes ; spherically symmetric , 1-d , lagrangian , flux - limited code which uses pseudo - viscosity and determines opacities calculated for hydrogen , helium , and oxygen rich layers .
modifications included a geometric dilution model for gamma - ray deposition , and updated opacity tables @xcite .
rayleigh - taylor mixing and nuclear reactions are not included in the calculations .
the rosseland mean opacities are tabulated as functions of composition , temperature , and density , assuming local thermodynamic equilibrium .
two effects not included in the rosseland mean are line opacities and non - thermal excitation or ionization from gamma rays . in order to account for these
affects an opacity minimum ( opacity floor ) is set at 0.25 @xmath26 g@xmath27 for the helium rich core and 0.01 @xmath26 g@xmath27 for the hydrogen rich envelope . +
the lc calculation is done in two steps , first the hydrodynamics is calculated , then in a second step the gamma - ray deposition is included in the computation .
the prescription for absorbed gamma rays requires calculating a deposition function following sutherland and wheeler ( 1984 ) at interval times of 3.5@xmath28 sec from 0 to 400 days .
the deposition function calculates the diffusion of gamma - rays from the location they are produced to regions outside the ni distribution , including gamma - rays escaping the ejecta . in the first step one
obtains the time evolution of density and volume of the ejecta that is necessary to calculate the absorbed gamma rays .
the energy produced by the gamma rays is not enough to affect the hydrodynamics and thus is not necessary to be included in the primary run .
this saves computation time , avoiding a calculation of the deposition function at each time step .
+ the @xmath1ni mass is varied using 3 different masses of 0.035 , 0.07 , and 0.14 @xmath0 which are mixed to three different regions ; less than 0.3 @xmath0 , 6 @xmath0 ( he core ) , and 11 @xmath0 ( core plus half the envelope , 7 @xmath0 for model g ) . the ni distribution is a step function resulting in a uniform spherical distribution of @xmath1ni . as the ejecta becomes thin , the gamma rays deposit energy in a symmetric sphere outside the @xmath1ni distribution .
the mixing of @xmath1ni is artificial since we are not mixing the composition along with it , but the goal of this study is not to create a rigorous model , only to explore the major parameters that influence the lc .
however it has been shown that the mixing of h into the he layer does affect the lc as seen in sn1987a @xcite .
the energy produced by the decay of @xmath1ni and @xmath1co is @xmath29 where s is the amount of energy per gram per second produced from the ni - co - fe decay , t is time in days , @xmath30=8.8 days and @xmath31=113.6 days and the amount being deposited in the sn ejecta ( @xmath32 ) is @xmath33 where dep is the deposition function . in it is useful to introduce a gamma - ray `` photosphere '' ( hereafter gammasphere ) , similar to how the actual photosphere is defined .
this is taken to be @xmath34 = 2/3 , found by integrating the gamma - ray cross - section 0.06 @xmath26 g@xmath27 and the net electron mole fraction y@xmath35 .
+ the gamma rays are produced from the decay of @xmath1ni to @xmath1co , releasing photons with an average energy of 1.75 mev , and @xmath1co to @xmath1fe , releasing photons with an average energy of 3.61 mev , and positrons with an average kinetic energy of 0.12 mev .
+ the optical photosphere is defined to be where @xmath36=2/3 , integrating inward from the surface . the optical opacities are calculated and tabulated using bound - bound , free - free and bound - free transitions for multi - level atoms of h , he , c , o , si , fe . these tables were made for calculations where the h envelope was metal deficient , such as sn 1987a .
this is not a problem in lc calculations because in the region where the optical depth is high most of the opacity comes from electron scattering and the metal lines are an insignificant contribution . in the optically thin regions the opacity is so low that the contribution of metal lines to the opacity is small .
plus the metalicity of progenitors is unclear , sne found far from the nucleus in large galaxies with steep metalicity gradients most likely will have low metalicity envelopes .
+ the total luminosity ( l ) at any time is defined as the luminosity at the photosphere ( @xmath37 ) plus the energy produced by the deposition of gamma rays above the photosphere which is assumed to add to the bolometric luminosity .
@xmath38 where @xmath39 is the position of the photosphere , @xmath40 is the radius of the ejecta , @xmath41 is the mass at radius @xmath42 , and @xmath32 is energy per gram per second deposited by gamma rays at radius @xmath42 . at late times
when the gammasphere has receded through the ejecta the energy source of the lc is just the spontaneous release of energy deposited by the decay of ni - co - fe .
the absolute magnitude light curves are calculated following swartz et al .
( 1991 ) , @xmath43 where @xmath44 is the absolute bolometric magnitude and @xmath45 is the total luminosity in units of @xmath46 ergs .
visual and blue light curves are calculated by assuming a blackbody at the effective temperature integrated over the v and b response curve given in azusienis and straizy ( 1969 ) .
the photospheric temperature , t@xmath47 , is defined by @xmath48 following swartz et al .
( 1991 ) . where @xmath49 is the total luminosity and @xmath50 is the photospheric radius .
finally the velocity at the photosphere , v@xmath47 is the velocity of the material at the the photosphere .
the models a - h ( table i ) were constructed in order to easily compare the behavior of one parameter while holding the other parameters constant .
the standard model is model b ( m = 16 m@xmath6 , r = 430 r@xmath6 ) with e = @xmath23 ergs , m@xmath51 = 0.07 m@xmath6 , and ni mixing throughout the 6 m@xmath6 core .
these are similar to the values of sn 1987a except with a larger radius , taking into consideration that sn 1987a might have had a nonstandard radius .
thus i take the standard model as having the most average values around which the parameters are varied .
the 5 parameters are the progenitor radius , envelope mass , explosion energy , mass of @xmath1ni , and @xmath1ni mixing .
all lc graphs are plots of absolute magnitude versus time in days and all calculations proceed to 400 days and include the observed lc of sn 1987a for comparison @xcite .
1 - 3 show the affects of progenitor radius , figs .
4 - 6 show the affects of ejected mass , figs . 7 - 9 show the affects of explosion energy , figs .
10 - 12 show the affects of ni mass , figs .
13 - 15 show the affects of ni mixing . figs .
1 , 4 , 7 , 10 , and 13 show bolometric and visual light curves , and photospheric temperature and velocity for each of the `` affects of '' series . figs .
2 , 5 , 8 , 11 , and 14 , show the density and temperature profiles for each of the `` affects of '' series at times 0 , 36 hours , and 47 days .
also shown is the final velocity profile and the luminosity versus mass at times 94 , 189 , and 379 days .
3 , 6 , 9 , 12 , and 15 , show the affects on the lc at different values of the parameters that are held constant for each of the `` affects of '' series and for reference compare them to the instantaneous energy released from the ni - co decay . in this way i am exploring the parameter space in the most systematic way possible .
+ for example , the affects of radius graphs show models with the radius ranging from 43 r@xmath6 to 4300 r@xmath6 while holding the envelope mass at 16 @xmath3 , explosion energy at @xmath23 ergs , mass of @xmath1ni at 0.07 @xmath0 and the @xmath1ni mixing to 6 @xmath0 .
, 1b , 1c , and 1d show the affects of radius on the bolometric lc , visual lc , t@xmath47 , and photospheric velocity respectively . figs .
2a , 2c show the affects of radius on the density , temperature versus mass at t = 0 , 36 hours , and 47 days after explosion .
figure 2b shows the final velocity versus mass and figure 2d shows the luminosity versus mass at t = 94 , 189 , 379 days after explosion .
3a , 3b , 3c , 3d each show the affects of varying the progenitor radius when changing the value of one parameter ; ejected mass to 8 @xmath52 , the explosion energy to @xmath24 ergs , the @xmath1ni mass to 0.035 @xmath52 , and the @xmath1ni mixing @xmath53 0.3 @xmath3 , respectively .
a general overview of the lc can be explained as follows .
the lc gets its energy from two sources , deposited shock energy and deposited gamma rays .
almost always the early light curve ( @xmath53 50 days ) is powered by the diffusive release of internal energy deposited by the shock wave as it propagates through the envelope .
the middle lc ( 50 - 120 days ) , the plateau and secondary peak , is one or a combination of both energy sources .
this region makes a transition from being powered by deposited shock energy to deposited gamma - ray energy . and the late time light curve ( @xmath2 120 days ) , the tail , is powered by the instantaneous energy of deposited gamma rays . throughout the simulation
the luminosity is determined by the position of the photosphere calculated by integrating the opacity times the density from the surface inward .
as the model expands the opacity and density fall and the photosphere moves inward .
this movement of the photosphere toward the center of the model is called the recombination wave ( rw ) .
+ in brief : * larger progenitor radii produce a brighter early lc , and a longer and brighter plateau / secondary peak * the progenitor radius has no affect on the late - time light curve * the sn light curve of the smaller progenitor radii are dramatically influenced by the affects of ni the progenitor radius has a significant affect on the early lc and little or no influence at all on the late lc .
the light curve begins when the shock wave hits the surface .
once the shock breaks out the progenitor radius is the primary variable in determining the luminosity .
larger radii progenitors are already pre - expanded , have a large surface area , and produce a very bright peak ( fig .
1a , b ) . at late times
the initial radius has little affect on the gamma - ray deposition .
+ the initial delay in the light curve larger radii progenitors is due to the shock taking longer to arrive at the surface .
the shock wave can be identified in figure 2 by the deviation of the long dashed lines and the solid lines for each of the three models .
the time the shock wave spends inside the largest radius progenitor is 8 days compared to 2 hours for the smallest radius .
it can be seen in figs . 2a , and 2c that the shock in model a has already reached the surface by t = 4.2 hours , while both models b and c have shocks moving down the density gradient near 10 m@xmath6 and 7.5 m@xmath6 , respectively . in figure 2c
the temperature profiles clearly show a forward shock and a reverse shock forming an mesa in the middle . in the largest radii model
the deposited shock energy will not do as much pdv work , since it is already in an expanded state when the shock arrives , and the material will stay hotter longer .
once the shock wave emerges for models with increasingly larger radii the radiation has a smaller diffusion time and the photon flux is higher accounting for the progressively broader initial peak and plateau .
a comparison of the diffusion times between the three models clarifies this ; the ratio of model a to b is 10 , model a to c is 100 using diffusion time scale at constant opacity @xmath54 3@xmath55m/(4@xmath56rc ) .
+ the initial peak of model c has a 4 day rise time compared to the almost instantaneous rise time of models a and b ( see fig .
1a , 1b , 1c , 2d ) . this can be explained by the behavior of the shock wave in low , constant density material .
as the shock nears the surface of model c it becomes smoothed out and looses it
s `` shock '' definition and more mass has a higher temperature , giving a broad peak to the photosperic temperature evolution .
this is unlike the smaller radii models which have a sharper , more defined shock and thus achieves a higher temperature , but then expands and cools faster .
+ the difference between the radii can explain the maximum lc luminosity even though the photospheric temperature in the largest radii model is lower .
the ratio of the largest to smallest radius model is 100 , while the ratio of the breakout temperature is only about 2.5 ( fig .
this leads to a factor of about 256 brighter in luminosity ( 6 in magnitude , figs . 1a , 1c , 1d ) for the larger radius progenitor .
for smaller stars much more of the internal energy deposited by the shock is transformed into pdv work in order to unbind the star .
the initial burst is much brighter , seen as a spike , but falls rapidly with the expansion .
+ at late times the luminosity profiles are a useful diagnostic for the gamma - ray deposition .
all three models are shown at t = 94 days ( model a long dash , b dotted , c small dash ) , and have indistinguishable luminosity profiles at t = 189 ( solid lines ) , and 379 ( dark solid lines ) .
the curves at t = 189 and 379 days are understandable since once the ejecta is in homologous expansion the velocity and density ( figs .
2a , 2b ) profiles have similar shapes and thus give similar gamma - ray depositions . the absolute magnitude and slope of the lc tails ( figs .
1 and 3 ) are similar since it is the mass of ni and the expansion velocities which are important in the deposition of gamma rays .
+ the visual light curve ( fig .
1c ) is shown for comparison to observations in the v band .
the radioactive tails all have a similar shape to figure 1a but reduced in brightness .
this shift is just due to the temperature being constrained , a temperature floor , when convolving with the v band filters .
the temperature floor is set at 4500 k. the photospheric velocity ( fig .
1d ) for the different models show the largest differences quite early and converge at about 100 days .
after day 35 the v@xmath47 is systematically higher for the larger radius model because the ejecta is evolving slower and thus the photosphere recedes inward at a slower rate . in the smallest radius model
it is interesting that the increase in the photospheric temperature during the secondary peak does not influence the photospheric velocity .
this is due to the photosphere residing in the he core and at temperatures of about 6000 k can not ionize the he . + the recombination wave is an indictor of the dynamics of the temperature and density of the different models . to gauge the dynamics
it is instructive to list the times of when each model is completely through the h envelope ; smallest to largest progenitor radius , 42 , 70 , and 104 days ( vertical arrows in figure 1b ) .
these models reach the center of the model , completely through the he core , at 60 , 103 , and 113 days , respectively ( vertical lines in figure 1b ) .
figure 3a shows that increasing the explosion energy has an increasing affect on larger radii progenitors .
the early lc gets progressively brighter as the progenitor radius is increased because the rw moves inward more rapidly and more of the shock energy gets radiated faster .
thus the characteristic features of the lc , the peak , plateau , and onset of the radioactive tail , are seen at earlier times .
after 200 days the material has expanded enough that gamma rays have started to escape and the lc begins to show a steeper slope due to the increased velocity profile of each model being higher than for models presented in figure 1a ( see affects of energy section ) .
+ figure 3b shows that placing all the ni at the center delays the affects of the deposited gamma rays and effectively traps more gamma rays at later times .
this has a pronounced affect on the smaller radius progenitors since most of the plateau and secondary peak is powered by @xmath1ni .
thus the plateau dips to even lower luminosities to indicate even more dramatically where the change from shock energy to deposited gamma rays occurs .
the secondary peak then appears later , flatter , and fainter .
+ figure 3c shows that reducing the mass of the h envelope produces a light curve that shows features like the onset and duration of the plateau , at earlier times . because of a lower h mass the photosphere can move more quickly through the envelope , producing a brighter and faster light curve .
the velocity , e / m , is faster than in figure 1a and thus the slope of the tail is steeper .
the discrepancy of the tails at late times is due to the velocity profile of the he core ( fig . 2b ) .
the 8 m@xmath6 models with only 2 m@xmath6 of h envelope have a larger percent of the mass with different velocity profiles than the 16 m@xmath6 models .
this is the reason why the spreading of the tails was not seen in the case where the explosion energy was increased to @xmath24 ergs ( fig .
3a ) . in that case
the velocity profiles were just scaled to higher velocities ( also see affects of mass section ) .
+ figure 3d shows that the ni mass affects the smaller radius progenitors more , reducing the plateau dip slightly and scaling down the entire secondary peak .
the largest radius model is unaffected by the change during the initial peak , plateau , and secondary peak .
however the luminosity drops further after the secondary peak to meet the tails of the other models .
the tails also have the same slope as figure 1a only scaled down in absolute magnitude by about 1 magnitude . in brief
: * increasing the mass of the h envelope causes all the lc features ( plateau , secondary peak , and tail ) to appear at later times * increasing the h envelope mass produces a fainter early lc and a slowly declining radioactive tail * reducing the h envelope mass changes the final density profiles causing the lc tail to be very steep the general trends seen in figures 4 - 6 show that a more massive envelope results in a fainter early lc , a more pronounced plateau and secondary peak , and a slower decline of the tail .
the affect is due to the significant difference in the velocity profiles ( fig .
increasing the ejected mass and keeping the explosion energies the same lowers the energy per gram , thus the velocity , and the characteristic features of the lc are seen at a later time , a slower lc is produced .
+ the early lc is dependent on the mass because a larger mass envelope will have a longer diffusion time and thus trap the radiation longer and results in a fainter peak . during the plateau region
a higher mass h envelope will take longer to expand and cool as seen by the different positions of the rw between the models at day 47 ( figs .
5a and 5c ) .
the position of the photosphere at 47 days ( figure 5c ) is at the top of rw , located at 12 m@xmath6 for model b , 8 m@xmath6 for model e , and 6 m@xmath6 for model g. in the largest mass model the rw has more mass to move through so the photosphere will take longer to arrive at the he core , accounting for the longer plateau duration .
figure 4b supports this showing that the photospheric temperature for larger ejected mass stays hotter for a longer time .
the smaller the mass of the envelope the higher the velocity profile ( fig .
5b ) , the lower the density profile ( fig .
5a ) , the lower the temperature profile ( fig .
5c ) , the faster the rw and thus little time for the plateau to form .
since the late time lc is dependent on the trapping of gamma rays the smaller mass explosions will start losing gamma rays sooner for two reasons .
first the number of absorbers is reduced and second the density profile is lower at any given time .
thus the lc tail will have an increasingly steeper slope as seen in figs .
4a , 4c , and 4d .
figure 4a shows the affect of an extreme case of having no hydrogen envelope .
this model has only the 6 m@xmath6 he core , with its original unmodified radius .
the ni mass and ni mixing are consistent with the other three models .
it can be seen that that the initial rise in brightness is reduced since the h envelope mass is absent and the progenitor radius is smaller .
the affect of removing the h envelope causes the lc to have an earlier secondary peak similar to that seen from explosions of wolf - rayet stars @xcite .
the secondary peak and the lc tail are not as bright compared to the other models due to the photophere receding faster into the center of the model and to the fewer absorbed gamma rays heating the material and adding to the luminosity .
the visual lc s ( fig .
4c ) show that for a smaller ejected mass the initial peak becomes more defined , whereas the higher mass models show a more extended plateau . the photospheric velocity ( fig .
4d ) for all three models have identical early fall times , showing that the expansion in the outer regions is very similar .
then after about 25 days they start to separate showing that as the photosphere reaches the he core and the photosphere recedes to much lower velocities faster .
for the 8 m@xmath6 model this occurs at 40 days , 12 m@xmath6 model at 65 days , and 16 m@xmath6 model it is 90 days .
the influence of the heating due to gamma rays can be seen in the luminosity profiles in figure 5d . at t
= 94 days ( solid lines ) the 8 and 12 m@xmath6 models have the same luminosity profiles indicating that the photosphere has receded into the most inner most material and the total luminosity is given by the trapping of gamma rays . due to its large mass the 16 m@xmath6 model can still release energy stored from the trapped gamma rays out past day 94 . at t = 187 days
( dashed lines ) the photosphere in all models has receded to the inner most mass zones . the 8 m@xmath6 model has had a movement of the gammasphere inward in mass indicated by the the lower luminosity profile .
this is evidence that the material has become thin to gamma rays and some are escaping . by t
= 379 days ( solid dark lines ) the luminosity profiles are diverging , showing that the smaller mass models have increasingly more gamma rays escaping .
+ figure 6a shows that for a higher energy the early lc is just scaled brighter due to the increased shock energy being released .
the plateau is shorter since the material is expanding faster and thus cooling faster and the recombination wave moves inward in mass faster .
the most pronounced affect is the fanning of the tails due to the increased expansion velocity and thus the escape of gamma rays .
this shows that increasing the explosion energy has an increasing affect on the escape of gamma rays for smaller masses .
+ figure 6b shows that with smaller amounts of ni the early lc is unchanged , but the luminosity drops near the end of the plateau , affecting both the secondary peak and tail .
the diffusive release of shock energy was not influenced by the reduction of ni mass so no change in the early lc or most of the plateau is seen .
the secondary peak is affected since it is partially powered by ni and figure 6c shows a reduction in its brightness and width .
the most significant affect is the tails which show the same slopes as in fig 3a but the absolute magnitude is greatly reduced .
+ figure 6c shows that for a larger radius the early lc is much brighter due to a larger radiating surface .
the duration of the plateau is increasingly longer for larger ejected masses .
however the width of the secondary peak does nt seem to be influenced at all .
this is due to the recombination wave moving through the he core which does nt have the required internal energy to sustain the higher luminosity .
thus it recedes fast releasing stored shock energy and deposited gamma rays .
the tail slopes and absolute magnitudes are exactly the same as for figure 3a ( see affects of radius ) .
+ figure 6d shows that confining the ni to the innermost mass layers ( @xmath53 .3 m@xmath6 ) has an increasingly larger influence on the smaller ejected masses . for smaller masses
the secondary peak is much longer in duration due to the appearance of the large dip at 25 days .
the dip is also found for larger masses but the affect is reduced .
the reason for the large dip is due to the change in the lc being powered by deposited shock energy to deposited gamma rays .
this transition between energy sources is a smooth transition , possibly due to the treatment of gamma - ray deposition .
other light curve studies have used the mixing of h and he to reproduce the same transition for sn 1987a ( woosley 1988 , shigeyama 1988 , utrobin 1991 ) .
the purpose of h / he mixing was to reduce the number of free electrons and thus expedite the recombination wave inward .
the photosphere then reaches the ni bubble faster and the affect on the light curve is similar to mixing the ni outward , showing a dip and a well defined secondary peak . in both scenarios
the photosphere rapidly reaches the ni bubble where the material is hot enough to cause an increase in the luminosity . at late times
the tails all show total trapping of the gamma rays .
the affects of mass shown here are consistent with woosley ( 1988 ) , shigeyama and nomoto ( 1990 ) , arnett ( 1989 ) , and hsu et al .
( 1992 ) when comparing both the luminosity differences in the secondary peak and time of the secondary peak maximum . in brief
: * higher explosion energies produce a lc with features ( onset and duration of plateau , secondary peak ) that occur at earlier times * higher explosion energies produce a bright early lc and a steeper slope of the tail * increasing the energy does not change the density profiles as significantly as changing the envelope mass ( section 5.2 ) , limiting the spread in the radioactive tails the general trends found when the explosion energy is varied can be seen in fig 7a . as the energy is increased the early lc and the plateau are scaled brighter and the tail has a steeper slope . increasing
the explosion energy produces a larger e / m , energy per gram , and thus a higher overall velocity profile .
the higher the explosion energy the more energy deposited , the higher the temperature and the brighter the early lc and plateau in most cases because they are powered by the release of deposited shock energy . the larger velocities expand and cool the material giving a faster rw and and a faster overall lc . at later times
the lc tails for a higher explosion energy have steeper slopes , but not as much of an affect as when varying the mass ( see affects of mass ) .
this can be understood from a comparison of velocity profiles . the velocity profile ( fig .
8b ) shows that the final velocities do not differ by nearly as much as compared with figure 5b ( varying the m@xmath57 ) .
furthermore the density profiles figure 8a and temperature profiles figure 8c show more similar profiles for different explosion energies than figures 5a and 5c . at late times
the `` affects of energy '' series of models have very similar density and velocity profiles in comparison to the `` affects of mass''series of models .
the similarity in velocity and density profiles accounts for the similarity in the lc tails in figure 7a .
the important result here is that given a certain e / m it does not necessarily mean varying e or m@xmath57 will result in the same lc tail . +
a comparison between changes in e or m@xmath57 also affects the photospheric temperature of the light curve . the photospheric temperature ( fig .
7b ) shows a much slower response to a variation in energy than the photospheric temperature when varying the mass ( fig .
4b ) . a comparison in the time difference of when the photospheric temperature drops to 4500 k between varying the energy and varying the mass shows this .
when the energy is tripled the time difference is only 27 days as opposed to when the mass is just halved the difference is 57 days .
thus the temperature of the photosphere is much more sensitive to a change in mass rather then a change in energy .
+ the visual lc ( fig .
7c ) shows very little or no change in the lc shape between differences in models when compared to figure 1a , except for the initial peak becoming part of the plateau .
the photospheric velocity for all models is relatively the same until the photosphere enters the he core region where the velocities separate .
the models with the highest energy show that they reach the he core the fastest and thus start to move into the slower moving material earlier .
the luminosity profiles figure 8d show that for the least energetic model ( long dashed lines ) the affect of the gamma - ray heating wave is pushed to later times .
the more energetic models have already passed through that stage by the same time period . by t
= 190 days the profiles are exactly the same ( solid lines ) indicating that the difference in energies hasnt changed the deposition of gamma rays . at t
= 379 days the differences in the deposition can start to be seen as the luminosity profiles start to separate slightly for the different models .
+ figure 9a shows that a smaller the ejected mass results is a smaller difference in the early lc and a steeper fall of the tail .
this is understandable since the smaller the h envelope the faster the recombination wave and the faster the release of internal energy deposited by the shock . at late times
the steeper tail can be explained by a higher overall velocity profile than figure 6b .
this is due to the lowered h mass , and progressively more gamma rays escape with increasing explosion energy .
+ figure 9b shows that when ni mixing is changed to @xmath53 0.3 m@xmath6 the early lc is exactly the same as in fig 1a .
but the plateau and secondary peak are systematically affected more .
there is a clear separation between the thermal and ni energy .
this is seen as a more defined plateau / secondary peak .
the tails are nearly the same since most of the gamma rays are absorbed .
figure 9c shows that changing the radius to 4300 r@xmath6 distinguishes between the lc peak times which are about 1.4 and 1.7 times earlier for the @xmath24 and @xmath25 erg models respectively .
the affects of a larger radius show a narrower range in the duration , slope , and absolute scale of the plateau as compared to the same graph comparing ejecta masses ( fig .
the secondary peak is smaller in width and starts to blend with the plateau .
this is due to more shock energy deposited into the h envelope and he core .
the tail is almost identical to figure 7a but with a greater change in slope with increasing energy .
again the influence is much smaller than with a variation in mass ( fig .
+ figure 9d shows that when the ni mass is decreased the secondary peak gets increasingly fainter but more importantly the the width of the peak gets smaller with increasing energy .
this means that as the explosion energy gets higher , the secondary peak is more powered by gamma - ray energy deposition , while the lower energy explosions have a longer diffusion time and the shock energy can still contribute to the secondary peak . at later times
the slope of the tails are similar to figure 7a but scaled to lower absolute magnitudes .
+ in brief : * larger @xmath1ni mass produces brighter radioactive tails and plateaus / secondary peaks * models that have smaller radii are influenced more dramatically by the @xmath1ni mass * larger e / m and extensive @xmath1ni mixing reveals the @xmath1ni affects earlier and steeper slope of the tail the general trend found when varying the @xmath1ni mass shows that as the @xmath1ni mass is increased the plateau , secondary peak and tail of the lc get brighter .
the early lc is powered by deposited shock energy so changing the ni mass should not affect this part of the light curve . by about 20 to 50 days the photosphere moves into the region where gamma - ray deposition is important .
the affects of ni mass are apparent by the difference in absolute magnitude and shape of the lcs . in figure 10a
the affects of an increased gamma - ray deposition can be seen by all three lc s diverging at @xmath54 40 days . as the ni mass
is increased the secondary peak increases in both absolute magnitude and width .
in fact when the ni mass is 0.14 m@xmath6 the secondary peak is brighter than the plateau , and in the v light curve the secondary peak is even brighter than the initial peak ( fig .
at late times the v light curve have similar tail slopes but scaled to lower luminosity ( equation 3 ) . for comparison a model
is shown that contains no ni mass and abrutly falls after the plateau .
since having no ni mass eliminates the secondary peak it is reasonable to assume that ni heating plays a role in producing the secondary peak ( fig 10a ) .
+ the plateau is actually lengthened by increased amount of ni as seen in all graphs to the point where it is almost doubled for 0.14 m@xmath6 ni . the time of the secondary peak maximum increases with increasing ni mass .
this can be explained by the ni - co decay keeping the material hotter for a longer time , thus slowing the rw .
figure 10b shows the photospheric temperature for the 0.14 m@xmath6 ni model stays hot for greater than 115 days , indicating that the gamma rays participate in the heating of the material below the photosphere .
there is little change in the photopspheric velocity ( fig .
10d ) which shows no difference until about 55 days .
the slight difference in velocity after 55 days is due to an increase in the opacity , causing the photosphere to move into faster moving material .
the slopes of all the tails ( figs .
10a , 10c ) are similar since there is no variation in mass , energy , or ni mixing .
this can be easily explained since all models have almost identical density , velocity , and temperature profiles figs .
11a , 11b , 11c . in figure 11c the heating due to ni can be seen to slightly affect the temperature profile at day 47 in the region @xmath53 6 m@xmath6 .
figure 11d shows the most predominate affect of ni mass .
the luminosity profile directly reflects the amount of energy supplied by radioactivity , similar profiles but different absolute luminosities . in general since the only source of energy at lates times is the decay of co it is possible to estimate the ni mass based on the absolute magnitude of the tail .
however , a lc tail with a steeper slope than the decay rate of co indicates that gamma rays are escaping the ejecta and thus an under - estimate of the ni mass would be obtained .
+ figure 12a shows that lowering the h envelope mass enables ni to power the full plateau .
the affects of ni appear at day 20 due to the recombination wave moving quickly through the low mass h envelope and uncover the regions heated by gamma rays faster . as the ni mass is increased the he core is kept hot enough to allow the rw to move more slowly . at later times
the tails fall faster , when compared to figure 10a , due to the faster velocity , low density and thus less trapping of gamma rays .
+ figure 12b shows that confining the ni to @xmath53 0.3 m@xmath6 delays the affects of the ni energy source and consequently the lc continues to fall 15 days longer than the other models in figure 10a .
this is due to the ni taking longer to diffuses through more material .
it also produces a broader and longer secondary peak which also lengthens the plateau .
this is due to the ni being trapped longer and keeping the material hotter and thus keeping the rw from moving inward too quickly .
+ figure 12c shows that increasing the radius lets the deposited shock energy power the lc for a longer time , since the affects of the ni mass are nt seen until 80 days ( whereas for fig .
10a it is 40 days ) .
varying the ni mass does nt have as much of an affect on the plateau or the secondary peak because the rw travels inward slower due to the longer diffusion time
. just befor the secondary peak the photosphere moves through the he core the lcs start to diverage at about 140 days , showing that the model with the largest ni mass stays brighter for a longer time .
+ figure 12d shows that increasing the explosion energy , like decreasing the mass , lets the gamma rays power the lc earlier and lets the gamma rays escape at later times , so the lc falls below the instantaneous co slope .
again the affects of increasing the energy are not as dramatic as lowering the h mass by the same factor . in brief
: * extensive @xmath1ni mixing causes the ni heating to be seen only slightly earlier and the plateau to be slightly brighter * confining the @xmath1ni delays and enhances the heating when the photosphere reaches the ni - powered diffusion wave ( ni bubble ) * larger @xmath1ni mixing causes only a slightly steeper radioactive tail the general trends found when the ni is mixed further out in the ejecta the earlier the affects of the gamma rays are seen and at late times the slope of the tail becomes steeper .
however the affects of ni mixing are not as dramatic as varying the other parameters .
figure 13a shows that if ni is mixed further out the gamma rays have a longer path length and deposit their energy at greater distances .
so as the photosphere recedes , the more mixed models show the influence of the gamma rays earlier , see figs .
13a , 13b , 13c .
it is expected that as mixing of ni is varied there should be a time difference between models as to when the ni starts to influence the light curve .
the model with the greatest ni mixing ( 11 m@xmath6 ) has a lc that becomes brighter earlier , by about 20 days , than the other two less mixed models .
this is expected since the photosphere will encounter the gamma - ray heated material faster in the model with greater ni mixing .
+ the more mixed models have a brighter and slightly broader plateau due to a combination of two things . at earlier times
the plateau is brighter because more gamma - ray energy depostions can heat the material nearer to the photosphere adding to the luminosity . secondly gamma - ray deposited energy outside the photosphere adds directly to the luminosity .
the photospheric temperature of the more mixed models is larger but not enough to account for the difference in brightness on the plateau , thus the additional luminosity is due to gamma ray deposition outside the photosphere .
the secondary peak becomes broader as the ni mixing is more confined to the central region .
this is due to the slow diffusion of material being heated by the gamma rays , the ni bubble or ni - powered diffusion wave .
the photosphere can be seen to increase in temperature as the rw reaches the ni bubble in figure 13b and is shown to be slightly earlier for more mixed models . for the more mixed model this produces a wider plateau that merges with the secondary peak . for the least mixed model the plateau ends earlier and
produces a dip in the light curve , resulting in a more defined secondary peak ( figs .
13a , c ) . overall the ni mixing is not a large factor in differentiating models . in figure
14d the more mixed models have a luminosity profile that increases with increasing mass , showing the location of the ni distribution . at later times
the more mixed models have gamma rays escaping , as seen by a scaling down of the luminosity profile , indicating that the gammasphere is moving inward faster . the density , temperature or velocity profiles ( figs .
14a , 14c , 14d ) show very little difference for all models out to 400 days .
+ figure 15a shows that for a smaller ejected mass the ni mixing between the core ( dotted line ) and the envelope ( dashed line ) is not significant , which is expected since it is only a difference in mixing of 1 m@xmath6 .
however the least mixed model ( solid line ) shows a dramatic early drop in luminosity with very little plateau . as the photosphere recedes into the he core material that has been heated by the trapped gamma rays it produces a secondary peak similar to sn 1987a . in the least mixed model since
all the gamma rays are trapped the radioactive lc tail follows the instantaneous co decay slope .
+ figure 15b shows that when the amount of ni is reduced the plateau drops earlier and the light curves look more similar to each other .
compared to figure 13a the secondary peak arrives earlier , has a lower absolute magnitude and a shorter width . and
as expected the difference in the tails is similar to figure 13a but scaled fainter .
these light curves represent the most severe case of no affect on the lc .
+ figure 15c shows that with an increased progenitor radius the mixing of ni has very little affect on any part of the light curve .
+ figure 15d shows that when the explosion energy is increased the early lc become both more similar and brighter , and the tails tend to diverge only slightly more than figure 13a .
this indicates that at higher energies ni mixing is even less important .
a numerical parameter study of light curves showing the affects of five parameters representing the progenitor and explosion is carried out to 400 days . in this parameter study
it is found that the early lc is affected by r , m , and e ; the plateau is affected by r , m , e , m@xmath51,and @xmath1ni mixing ; the secondary peak is affected by the r , m , e , m@xmath51 , and @xmath1ni mixing ; and the tail is affected by m , e , m@xmath51 , and @xmath1ni mixing .
it was found that the primary parameters that influence the overall behavior of the lc are the progenitor radius and the ejected mass .
these are the two parameters that give the largest changes in the lc with reasonable variations in the parameters .
the secondary parameters are the explosion energy , ni mass , and ni mixing .
the ejected mass was determined a primary parameter since the ejected mass is thought to vary from 4 m@xmath6 to up 25 m@xmath6 and affects on the light curve sufficiently account for the observed lc variation .
the explosion energy is taken as a secondary parameter since the explosion mechanism is not well determined and is thought to span a smaller range in parameter space for normal type iis , from 0.5 to 2@xmath7 ergs . + it is possible that a given lc is not unique since many variables affect the shape and absolute magnitude of the light curve
. however it is more plausible to say that there is a small range in parameter space where the lc is not unique .
it is then necessary to look at the entire lc ( out to 400 days or later ) in order to determine accurate values of the parameters .
it is also beneficial to have a consistent model by fitting the spectrum in conjunction with the light curve , similar to the analysis of sn 1997d by @xcite .
+ the progenitor radius has the largest affect on the early lc but it was also varied by the largest factor , 100 compared to a factor of 3 for the explosion energy .
the radius is the one parameter that can reasonably explain the range in peak absolute magnitudes of the light curve .
+ the conspicuous secondary peak found in almost all light curves just after or in place of the plateau is due to the recombination wave moving quickly through the compositionally unmixed he core and reaching the area where gamma - ray energy was deposited .
+ the influence of @xmath1ni becomes increasingly important for stars with smaller radii and smaller ejected mass or larger explosion energy .
but @xmath1ni has its most significant affect on a smaller radius as demonstrated with sn 1987a .
thus as the progenitor radius is decreased the plateau and secondary peak become powered by the deposition of gamma rays .
the affect of ni mixing on the light curve can dissapear if the progenitor radius is large or the amount of ni in the ejecta is very small .
+ the variation of ejected mass does not necessarily show inverse affects when compared to changing the explosion energy .
this is especially true for the plateaus , secondary peaks , and tails .
it is shown that changing the h envelope mass causes differences in the density and velocity profiles that change the light curve more dramatically than the explosion energy .
thus it would be expected that if the he core and the h envelope mass were changed in proportion the inverse affects should appear .
it is also expected that if polytropes were used then changing the envelope mass would give more similar inverse affects .
+ i would like to sincerely thank eddie baron and david branch for resources , guidance , and necessary discussions , robert harkness for the original hydrocode and opacity tables , and stan woosley for the numerically evolved stellar models .
i am grateful for the referee whose extensive comments and suggestions greatly improved the paper .
this work was supported under nsf grant eps-0132289 and nasa grant ncc5 - 399 .
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx = a @xmath9 16 1,2 0.035,0.07 center , core + b @xmath10 16 1,2,3 0.035,0.07,0.14 center , core , env .
+ c @xmath11 16 1,2,3 0.035,0.07,0.14 center , core , env .
+ d @xmath10 12 1,2 0.035,0.07,0.14 center , core + e @xmath11 12 1 0.07 core + f @xmath9 8 1 0.07 core + g @xmath10 8 1,2,3 0.035,0.07,0.14 center , core , env . +
h @xmath11 8 1 0.07 core + | results of numerical calculations of type ii supernova light curves are presented .
the model progenitor stars have 6 @xmath0 cores and various envelopes , originating from a numerically evolved 20 @xmath0 star .
five parameters that affect the light curves are examined : the ejected mass , the progenitor radius , the explosion energy , the @xmath1ni mass , and the extent of @xmath1ni mixing .
the following affects have been found : 1 ) the larger the progenitor radius the brighter the early time light curve , with little affect on the late time light curve , 2 ) the larger the envelope mass the fainter the early light curve and the flatter the slope of the late light curve , 3 ) the larger the explosion energy the brighter the early light curve and the steeper the slope of the late light curve , 4 ) the larger the @xmath1ni mass the brighter the overall light curve after 20 to 50 days , with no affect on the early light curve , 5 ) the more extensive the @xmath1ni mixing the brighter the early light curve and the steeper the late light curve .
the primary parameters affecting the light curve shape are the progenitor radius and the ejected mass .
the secondary parameters are the explosion energy , @xmath1ni mass and @xmath1ni mixing .
i find that while in principle the general shape and absolute magnitude of a light curve indicate a unique set of parameters , in practice it is difficult to avoid some ambiguity in the parameters .
i find that the nickel powered diffusion wave and the recombination of helium produce a prominent secondary peak in all our calculations .
the feature is less prominent when compositional mixing , both @xmath1ni mixing and mixing between the hydrogen and helium layers , occurs .
the model photospheric temperatures and velocities are presented , for comparison to observation . |
the cosmological principle is one of the most fundamental assumption in modern cosmology which states that the universe is statistically homogeneous and isotropic on sufficiently large scales .
homogeneity implies translational invariance i.e. the universe looks same to all observers at different locations .
isotropy implies rotational invariance i.e. the universe looks the same in all directions .
these two assumptions play an important role in the analysis and interpretation of various cosmological observations .
one can not prove these assumptions in a rigorously mathematical sense but it is possible to verify them using various observations .
the near uniform temperature of the cosmic microwave background radiation is considered to be one of the best possible evidence in favour of isotropy @xcite .
there are many other observations in favour of isotropy such as the angular distributions of radio sources @xcite , the x - ray background @xcite , the gamma - ray bursts @xcite , the distribution of galaxies @xcite , the distribution of supernovae @xcite and the distribution of neutral hydrogen @xcite .
but isotropy around us alone does not guarantee homogeneity of the universe because homogeneity and isotropy may or may not coexist .
the universe can be isotropic around a point without being homogeneous .
the homogeneity can be inferred from isotropy only when there is isotropy around each and every point in the universe .
so it is not quite straightforward to infer homogeneity of the universe from isotropy around us alone .
the statistical properties of galaxy distributions can be characterized by the correlation functions @xcite .
the two point correlation function quantifies the clustering strength which is well described by a power law on small scales and vanishes on large scales conforming to large scale homogeneity .
however a major drawback of the correlation function for the analysis of homogeneity arise from the fact that it assumes a mean density on the scale of the survey which is not a defined quantity below the scale of homogeneity .
it is well known that the galaxy distribution behave like a fractal on small scales . but @xcite analyzed the cfa1 survey and suggested that the distribution of galaxies is fractal to arbitrary large - scales .
further , @xcite supported the argument of fractal universe by analyzing different samples .
there are other studies which claim the absence of any transition to homogeneity out to scale of the survey @xcite .
@xcite claimed that the fractal structure is valid at small scale but at large - scale the universe is homogeneous .
@xcite analyzed the perseus - pisces redshift survey and claimed that galaxies are clustered at small scale and intermediate scale but is homogeneous on large - scale .
@xcite found the volume - limited subsample of ssrs2 to be consistent with both fractality at small scale and homogeneity on large - scale .
@xcite find homogeneity at scale larger than 80 @xmath7 , using multifractal analysis of the lcrs sample .
@xcite used fractal analysis to pscz and find the homogeneity scale even at 30 @xmath7 .
@xcite have analyzed the two - dimensional strips from sdss dr1 using multifractal analysis and found the transition to homogeneity occurring at length - scale @xmath8 @xmath7 .
@xcite have analyzed the distribution of sdss luminous red galaxies ( lrg ) to find the transition to homogeneity at @xmath9 @xmath7 .
@xcite used the multifractal analysis of sdss dr6 to find the transition to homogeneity at length - scale greater than @xmath10 @xmath7 .
@xcite find homogeneity at length - scale @xmath11 @xmath7 in the redshift range @xmath12 using wigglez survey .
@xcite find homogeneity at length - scale above @xmath13 @xmath7 for the sdss dr7 quasar catalogue containing the huge - lqg at redshift @xmath14 .
some recent studies with shannon entropy show that the main galaxies from sdss dr12 and luminous red galaxies ( lrg ) from sdss dr7 are homogeneous beyond @xmath15 @xcite .
as there is no consensus in the issue of cosmic homogeneity , it is important to test the assumption on multiple datasets with different statistical tools .
if the assumption is ruled out with high statistical significance by multiple datasets there would be a major paradigm shift in cosmology .
sdss is the largest and finest galaxy redshift survey todate which provides us an unique opportunity to test the assumption of cosmic homogeneity on very large scales with unprecedented confidence .
sdss provides distributions of different types of galaxies out to different distances covering enormous volumes .
galaxies are believed to be a biased tracer of the underlying mass distribution and different types of galaxies are expected to have different values of linear bias . in the present work we would like to test the assumption of homogeneity using different tracers of the underlying mass distribution .
we use the multifractal analysis @xcite to characterize the scale of homogeneity in the main galaxy sample , lrg sample and quasar sample from the sloan digital sky survey ( sdss ) @xcite .
we have considered the main galaxy sample ( mgs ) from sdss dr7 , luminous red galaxy ( lrg ) sample from sdss dr7 and quasars sample from sdss dr12 .
the lrg sample covers a much larger volume as compared to the main galaxy sample but it is only quasi volume limited and quite sparse as compared to the main sample .
the quasar sample analyzed here is even larger in volume .
it is the largest among these samples but also sparsest among them .
these samples provide us the scope to test homogeneity upto different length scales due to their different volume coverages . besides the different volumes and number densities
, the samples also have different properties such as luminosity and colour resulting in different clustering properties . a combined analysis of the multiple tracers of the underlying mass distribution using the same statistical tools ,
provide us an unique opportunity to test the assumption of cosmic homogeneity for different tracers on a wide range of length scales . a brief outline of the paper follows . in section 2
we describe the data followed by the method of analysis in section 3 .
we present the results in section 4 and conclusions in section 5 .
the sloan digital sky survey ( sdss ) @xcite is a wide - field photometric and spectroscopic survey of sky using a @xmath16 m sloan telescope at apache point observatory in new mexico , united states .
it imaged the sky in five different pass - bands @xmath17 , @xmath18 , @xmath19 , @xmath20 and @xmath21 .
we have considered three different types of tracers namely the galaxies from the main sample , the luminous red galaxies ( lrgs ) and the quasars from the sdss .
we have used @xmath22 , @xmath23 and @xmath24 throughout the analysis .
the various data sets used in the present analysis are described below .
we have used the main galaxy sample from sdss data release 7 @xcite for which the target selection algorithm is described in @xcite .
the main galaxy sample comprises of galaxies brighter than limiting r - band petrosian magnitude @xmath25 .
we have identified a contiguous region in the northern galactic cap which spans @xmath26 and @xmath27 , where @xmath28 and @xmath29 are survey coordinates defined in @xcite .
we have constructed a volume limited sample of galaxies in the region by restricting the r - band absolute magnitude to the range @xmath30 .
this absolute magnitude cut produces a volume limited sample of galaxies within @xmath31 .
we finally extract @xmath32 galaxies in the redshift range of @xmath33 which corresponds to the radial comoving distances in the range @xmath34 .
the luminous red galaxy distribution ( lrg ) @xcite from the sdss extends to a much deeper region of the universe as compared to the sdss main galaxy sample as they can be observed to greater distances as compared to normal galaxies for a given magnitude limit .
the lrgs have stable colors which make them relatively easy to pick out from the rest of the galaxies using the sdss multi - band photometry . for the present analysis
, we consider the spectroscopic sample of lrg extracted form the sdss dr7 .
the lrg sample is quasi - volume limited to a redshift of @xmath35 .
we use the dr7-dim sample extracted by @xcite .
this sample contains @xmath36 galaxies within the redshift range of @xmath37 to @xmath38 for g - band absolute magnitude ( @xmath39 ) range brighter than @xmath40 and fainter than @xmath41 .
we restrict our sample to a contiguous region in the northern galactic cap which spans @xmath42 and @xmath43 which constitutes our sample with @xmath44 lrgs .
the quasars are the brightest class of objects known in the universe and their high luminosities allow us to detect them out to larger distances .
we use the sdss dr12 quasar catalogue for which the target selection is described in @xcite and the data is described in @xcite .
the sdss dr12 catalogue contain a total @xmath45 quasars . in the present work we use a quasar sample prepared by @xcite .
the preparation of the primary quasar sample is described in @xcite .
the primary quasar sample used in this analysis contain @xmath46 quasars with the g - band psf magnitude @xmath47 and the r - band psf magnitude @xmath48 @xcite .
we select the quasars in a contiguous region @xmath49 and @xmath50 where @xmath51 and @xmath52 are the right ascension and declination respectively .
the selected quasars have redshift in the range @xmath53 . the resulting quasar sample have a varying number density in redshift . constructing a strictly volume limited sample of quasars from this data results into a sample with a very poor number density .
we construct a quasar sample with near uniform and reasonable number density in the entire redshift range by using a redshift dependent magnitude cut @xmath54 in the i - band absolute magnitude .
the limiting magnitude @xmath55 is described by a polynomial of the form @xmath56 where @xmath57 and @xmath58 are the coefficients to be determined .
we constrain @xmath59 so as to have a @xmath60 fluctuations in the comoving number density around the mean density at all the redshifts in the chosen redshift range .
we find that @xmath61 provides us with a quasar sample consisting of @xmath62 quasars distributed across a volume of @xmath63 @xmath64 with a mean number density of @xmath65 .
the quasar sample has a linear extent of @xmath66 in the radial direction .
large suite of dark matter simulations ( lasdamas ) ( mcbride et .
al . , in preparation ) is a cosmological n - body dark matter simulation project to obtain adequate resolutions in many large boxes rather than one single realization at high resolution .
the cosmological parameters used in the lasdamas simulations are those of @xmath67cdm model : @xmath68 , @xmath69 , @xmath70 , @xmath71 , @xmath72 and @xmath73 .
the galaxy mocks are generated by artificially placing galaxies inside dark matter halos using a halo occupation distribution with parameters fit from the respective sdss galaxy samples .
we have used the gamma release of the galaxy mock sample generated from oriana simulations .
the simulation was carried out in a simulation box of side @xmath74 with @xmath75 particles .
the individual particles in the simulation has a mass of @xmath76 and the spatial resolution of the simulation is @xmath77 @xmath78 .
we have used @xmath79 and @xmath80 independent realizations of mock galaxy catalogue from lasdamas simulations for the sdss main and lrg sample respectively .
we then extract mock samples for each of our sdss main sample and lrg sample which have the same geometry and number density as the actual data .
the mocks were analyzed exactly in the same way as the sdss data . with radius @xmath0 for the sdss main galaxy sample .
the @xmath81 errorbars are obtained from @xmath79 mock samples.,scaledwidth=35.0% ] [ fig : mgs_t1 ] [ fig : mgs_t2 ] [ fig : mgs_m1 ] [ fig : mgs_m2 ] [ fig : mgs_b1 ] [ fig : mgs_b2 ] with radius @xmath0 for the sdss lrg sample .
the @xmath81 errorbars are obtained from 40 mock samples.,scaledwidth=35.0% ] [ fig : lrg_t1 ] [ fig : lrg_t2 ] [ fig : lrg_m1 ] [ fig : lrg_m2 ] [ fig : lrg_b1 ] [ fig : lrg_b2 ] with radius @xmath0 for the sdss quasar sample .
the @xmath81 errorbars are obtained from bootstrap resampling.,scaledwidth=35.0% ] [ fig : qso_t1 ] [ fig : qso_t2 ] [ fig : qso_m1 ] [ fig : qso_m2 ] [ fig : qso_b1 ] [ fig : qso_b2 ] we have generated random samples for each of our data sets .
the random distributions have the same number density and they are distributed over the same region having identical geometry as the actual samples . in each case
we generate @xmath82 independent realizations of random samples .
the random samples were analyzed in the same way as the actual samples .
we have used two statistical measures to study the transition to homogeneity .
one is based on the simple number counts in spheres of radius @xmath0 and its scaling with @xmath0 .
the other one is the multifractal analysis which uses the scaling of different moments of the number counts .
one of the simple test of homogeneity is to check how the number of points included inside spheres changes with the radius of the spheres . for this
, we consider the galaxies as centres and place spheres of radius @xmath0 around these centres to count the number of galaxy @xmath83 within these spheres .
we consider only those galaxies as centres for which the spheres lie completely within the survey region .
the number counts @xmath83 is expected to scale as @xmath84 for a homogeneous distribution , where @xmath85 is the ambient dimension . in three dimension
we expect @xmath86 at the scale of homogeneity .
random distributions are considered to be homogeneous by construction .
we take the average over all the galaxies , to obtain the mean @xmath83 for a distribution and define an estimator , @xmath87 where @xmath88 and @xmath89 are the average number counts for sdss galaxy sample and @xmath90 random samples , @xmath91 is the ratio of number of galaxies in the @xmath92 random sample and sdss sample and @xmath93 is the number of random catalogues used in the analysis . it may be noted that the number of galaxies in the random sample and the sdss sample may not be identical due to the exclusion of spheres near the survey boundary .
we expect the value of @xmath94 to be unity at the transition scale to homogeneity .
galaxy surveys suggest that the universe resembles a fractal on small scales .
the fractal dimensions are commonly used to characterize fractals .
there are various ways to measure the fractal dimension @xcite , out of which the box counting dimension and correlation dimensions are simple to compute in finite distributions .
the box counting dimension and the correlation dimension quantify different aspects of the scaling behaviour .
these represents the particular cases of the generalized minkowski - bouligand dimensions @xmath95 , where @xmath96 represents different moments of galaxy counts .
the value of @xmath95 at @xmath97 and @xmath98 corresponds to box counting dimension and correlation dimension respectively . for a fractal distribution
, the values of @xmath95 will be different for different values of @xmath96 where @xmath99 gives more weightage to overdense region and @xmath100 gives more weightage to the underdense region .
the generalized correlation integral is defined as , @xmath101^{q-1},\ ] ] where @xmath102 is the number count within a sphere of comoving radius @xmath0 centered on the @xmath92 galaxy , @xmath103 is the number of centers chosen and @xmath104 is the total number of galaxies .
the generalized minkowski - bouligand dimension is given by , @xmath105 we have considered @xmath96 in the range @xmath106 . the finite number of galaxies restrict us to consider any arbitrary large values of @xmath107 @xcite .
for the calculation of @xmath95 from @xmath108 , we have used the numerical differentiation by riddle s method describe in @xcite .
the random samples are considered to be homogeneous by construction .
so to determine the scale of transition to homogeneity , we compare the results for various galaxy samples with that from the random samples .
if the galaxy sample is homogeneous then we expect the value of @xmath95 to be consistent with that of random samples at the transition scale to homogeneity .
figure [ fig : nvsrmgs ] shows the variation of the scaled number counts @xmath94 with @xmath0 for the main galaxy sample of the sdss .
the @xmath81 error bars are estimated from @xmath79 mocks samples .
we define the transition scale to homogeneity where the value of @xmath94 is within @xmath109 of its expected value @xmath110 .
the result suggests that the main galaxy sample is homogeneous beyond a length - scale of @xmath111 @xmath7 .
we show the correlation integral @xmath112 as a function of radius @xmath0 for @xmath113 and @xmath98 in the top two panels of figure [ fig : mgs_all ] . the solid line and the dot dashed line shows the results for the sdss main galaxy sample and its mock counterparts from the lasdamas n - body simulations and they are nearly indistinguishable from each other .
the result for the random samples is shown with the dashed line .
we find that @xmath112 increases with @xmath0 for positive values of @xmath96 whereas it falls progressively for negative values of @xmath96 .
it is quite clear from the plot that on large scales the main galaxy sample of sdss , the corresponding mock samples and the random samples all exhibit the same scaling behaviour .
but the scaling behaviours for the random samples are noticeably different from that of the sdss main galaxy sample and the mock samples at @xmath114 .
it may be noted that for both @xmath113 and @xmath98 these differences diminish with increasing length scales @xmath0 .
the behaviour of @xmath115 is similar for other positive and negative values of @xmath96 .
we show @xmath3 as a function of @xmath0 for @xmath113 and @xmath98 for the main galaxy sample in the middle two panels of figure [ fig : mgs_all ] .
the @xmath81 error bars for the sdss data are obtained from @xmath79 mock samples .
we expect @xmath3 to match with the ambient dimension for a homogeneous distribution .
the random samples are homogeneous by construction and we expect them to have @xmath116 at all @xmath0 .
we can see in both the middle panels of figure [ fig : mgs_all ] that for the sdss main galaxy sample and their mock samples @xmath3 values converge to @xmath117 within @xmath109 at scales @xmath118 . here
we adopt a working definition for the scale of homogeneity to be the scale where @xmath3 comes within @xmath109 of @xmath117 @xcite or @xmath109 of the results of the random samples whichever earlier .
it may be noted that @xmath119 values for the sdss and random samples differs greatly on small scales despite having the same mean inter particle separation .
this clearly indicates that the sdss main galaxy sample is homogeneous above a length scales of @xmath4 . in the left and right bottom panels of figure [ fig : mgs_all ]
, we show the variation of @xmath119 across the different @xmath96 values for @xmath120 and @xmath121 .
clearly at @xmath120 , the @xmath119 values for the main galaxy sample and its mock counterparts show a large offset from that observed in the random samples for the entire range of positive @xmath96 .
interestingly , increasing @xmath0 to @xmath122 reduces these offsets and the results for the random samples comes well within the @xmath81 errorbars of the results from the main galaxy sample .
this indicates the presence of a transition scale to homogeneity in the sdss main galaxy distribution at a length scale of @xmath122 .
figure [ fig : nvsrlrg ] shows the variation of measured number counts @xmath123 with r for the sdss lrg sample .
the @xmath81 error bars are estimated from @xmath80 mocks samples . following the definition stated earlier , we find that the lrg sample becomes homogeneous at length - scale of @xmath5 .
the top two panels of figure [ fig : lrg_all ] show the variation of @xmath124 with respect to @xmath0 for @xmath113 and @xmath98 .
we find that for @xmath99 , @xmath124 increases with @xmath0 whereas for @xmath100 , it falls progressively .
it is quite clear from the plots that the results from the mock samples are quite consistent with the lrg data on large scales .
but there behaviours are quite different as compared to the random samples at @xmath125 . in the middle two panels of figure [ fig : lrg_all ]
we show the variation of @xmath126 with @xmath0 for @xmath113 and @xmath98 for the sdss lrg sample .
the @xmath81 error bars for the lrg sample is estimated from @xmath80 mock samples .
the @xmath3 values for the lrg sample and its mock counterparts are quite different than the random samples at scale @xmath127 for @xmath113 and @xmath128 for @xmath98 .
following the same working definition mentioned earlier we find that the @xmath3 values for the lrg sample comes within @xmath109 of @xmath117 at a scale @xmath129 .
the variation of @xmath95 with @xmath96 for two different values of @xmath0 are shown in the bottom two panels of figure [ fig : lrg_all ] .
the bottom left panel in this figure corresponds to @xmath130 .
this plot clearly shows the transition to homogeneity is reached for the positive values of @xmath96 whereas the @xmath95 for negative values of @xmath96 still differs from the random samples indicating the non uniformity across the underdense regions at that scale .
the right panel shows the same results but for @xmath131 .
we find that the differences in the @xmath119 values at negative @xmath96 decreases noticeably and the errorbars for the lrg and random samples now overlap with each other .
clearly the spectrum of @xmath119 values for the lrg and random data agree well when we change @xmath0 from @xmath132 to @xmath133 .
this indicates a transition scale to homogeneity in the lrg data at @xmath6 . in figure
[ fig : nvsrqso ] we show the scaled number counts @xmath94 as a function of @xmath0 for the sdss quasar sample . the @xmath81 error bars shown here are estimated from samples obtained by bootstrap resampling .
we note that the scaled number counts for the quasar data does not show a distinct transition to homogeneity as seen earlier in the main galaxy sample and the lrg sample .
the quasar distribution seems to be quite consistent with a homogeneous distribution from small scales upto a length scale of @xmath134 .
but the scaled number counts show a small and near constant deviation from its expected value @xmath110 beyond @xmath135 .
this can not be emphasized due to a large mean inter - particle separation @xmath136 of the sdss quasar sample .
this departure does not necessarily imply that the quasar distribution is inhomogeneous and one needs to apply other statistical measures to asses its significance .
we show the variation of @xmath124 with @xmath0 for the sdss quasars for two different values of @xmath96 in the top two panels of figure [ fig : qso_all ] .
we find that for both @xmath113 and @xmath98 , apparently the value of the correlation integral @xmath124 for the quasar sample is quite close to that for the random samples throughout the entire length scale with small differences . in the middle left panel and right panels of figure [ fig : qso_all ]
we show the variation of @xmath119 with @xmath0 for the quasar sample for @xmath113 and @xmath98 respectively .
we find that the @xmath119 values for the quasar sample for @xmath98 are quite consistent with that from a random distribution for @xmath137 whereas the @xmath119 values differ from that a random distribution for @xmath113 upto a length scales of @xmath138 .
we show how @xmath3 depend on @xmath96 for the quasar sample for @xmath139 and @xmath140 in the bottom left and right panels of figure [ fig : qso_all ] respectively . clearly at @xmath141
the transition to homogeneity is observed for the positive values of @xmath96 whereas the @xmath95 for negative values of @xmath96 still differs from the random samples indicating the non uniformity across the underdense regions at that scale .
we also note that the @xmath119 values for both the quasar sample and the random samples deviate from @xmath117 for almost all the values of @xmath96 .
this may not be surprising as this length scale is very close to the mean interparticle separation in the quasar and random samples .
the right panel shows the same results but for @xmath142 .
we find that the differences in the @xmath119 values at negative @xmath96 decreases noticeably and the results for the random samples are well within @xmath81 errorbars for the results from the quasars .
further at @xmath140 the @xmath119 values converge to @xmath143 for positive @xmath96 values whereas the deviations still persists at negative @xmath96 values .
interestingly the deviations are of opposite sign for the quasar and the random samples .
@xcite find that the generalized dimensions @xmath119 for @xmath144 are extremely sensitive to the regions of low density for a finite size data sets .
this implies that one can not tell the difference between empty space and space that should be filled in with very low probability .
these differences dramatically affect the generalized dimensions @xmath119 for the negative @xmath96 values .
considering the very low density of the quasar sample we conclude that the sdss quasar distribution indicates a transition to homogeneity at @xmath5 .
we used the scaled number counts and different moments of the number counts in spheres to identify the transition scale to homogeneity in the sdss main galaxy sample , lrg sample and quasar sample .
the scaled number counts in the main galaxy sample and lrg sample show a transition to homogeneity on scales of @xmath4 and @xmath15 respectively whereas the scaled number counts in the quasar sample is consistent with homogeneity on small scales with a small and near constant deviation from it on larger scales . it may be noted here that the quasar sample is the largest but sparsest among all the samples we have analyzed .
further we could not define a strictly volume limited sample for the quasars as done for the main galaxy sample .
the lrg sample is a quasi volume limited in nature . to define our quasar sample we used a redshift dependent magnitude cut so as to maintain a near uniform comoving number density for them .
this helps us to maintain a uniform comoving number density in the quasar sample but introduces a non - uniformity in the magnitudes of the quasars in the sample .
further the resulting quasar sample has a very poor number density with mean inter - particle separation of @xmath145 and the number counts are expected to be dominated by shot noise below this length scale . given the incompleteness of the quasar sample it is hard to interpret the behaviour of the scaled number counts in the quasar sample in a simple manner .
we analyzed the spectrum of the generalized dimension for all three samples to find that the generalized dimension @xmath119 show a transition to homogeneity at @xmath122 , @xmath133 and @xmath15 for the sdss main galaxy sample , the lrg sample and the quasar sample respectively .
these values are somewhat different than that obtained with the scaled number counts of the respective samples .
this may arise due to higher sensitivity of @xmath119 to underdense regions .
one may note that these samples have quite different number densities .
the number density decreases by order of magnitudes as we go from main galaxy sample to the lrg sample and the quasar sample .
the transition scale to homogeneity are particularly different for the lrg and the quasar sample when they are determined from the @xmath96 dependence of the generalized dimension @xmath119 instead of the behaviour of their scaled number counts .
it is thus important to note that the transition scale to homogeneity is also sensitive to the statistical measures employed .
the present analysis indicates that there exist a transition scale to homogeneity for all the tracers of the mass distribution used in this analysis .
we find the transition scale to homogeneity to be quite different for the normal galaxies , lrgs and quasars .
this may originate from the fact that these distributions are biased tracers of the underlying mass distribution and the large scale linear bias for these samples are expected to be quite different from each other .
the quasars are found to inhabit dark matter halos of constant mass @xmath146 from the time of peak quasar activity ( @xmath147 ) and their large scale linear bias evolves from @xmath148 at @xmath149 to @xmath150 at @xmath151 @xcite .
our quasar sample extends from @xmath152 to @xmath153 for which we expect a large scale linear bias of @xmath154 . on the other hand
the sdss lrg sample is known to have a large scale linear bias values of @xmath155 @xcite .
so one would expect the quasar sample to be homogeneous on larger scales than the lrg sample .
@xcite applied an information theory based method @xcite to analyze the sdss dr12 quasar sample and find that the quasar distribution is homogeneous beyond a length scale of @xmath156 .
information entropy is related to the higher order moments of a distribution and may be more sensitive to the homogeneities present in a distribution .
but the present analysis shows that the sdss quasar sample is homogeneous on a length scale which is smaller than this and also smaller than the transition scale to homogeneity for the lrgs .
this indicates that the clustering strength as indicated by their bias values may not be the only parameter deciding the transition scale to homogeneity .
it should be also noted here that the quasar sample is very sparse .
the poor number density of the quasar sample together with its somewhat incomplete nature may also yield such differences in our results . however despite these differences all the galaxy samples from the sdss including the main sample , the lrg sample and the quasar sample exhibit a transition to homogeneity on large scales .
the present analysis thus validates the fundamental assumption of cosmic homogeneity and reaffirms that the universe is indeed statistically homogeneous on sufficiently large scales irrespective of the tracers .
b.p . would like to acknowledge financial support from the serb , dst , government of india through the project emr/2015/001037 .
bp would also like to acknowledge cts , iit kharagpur , iucaa , pune and tifr , mumbai for providing support for visit and use of its facilities for the present work .
bp thanks somnath bharadwaj for useful comments and discussions . | we carry out multifractal analyses of multiple tracers namely the main galaxy sample , the lrg sample and the quasar sample from the sdss to test the assumption of cosmic homogeneity and identify the scale of transition to homogeneity , if any .
we consider the behaviour of the scaled number counts and the scaling relations of different moments of the galaxy number counts in spheres of varying radius @xmath0 to calculate the spectrum of the minkowski - bouligand general dimension @xmath1 for @xmath2 .
the present analysis provides us the opportunity to study the spectrum of the generalized dimension @xmath3 for multiple tracers of the cosmic density field over a wide range of length scales and allows us to confidently test the validity of the assumption of cosmic homogeneity .
our analysis indicates that the sdss main galaxy sample is homogeneous on a length scales of @xmath4 and beyond whereas the sdss quasar sample and the sdss lrg sample show transition to homogeneity on an even larger length scales at @xmath5 and @xmath6 respectively .
these differences in the scale of homogeneity arise due to the effective mass and redshift scales probed by the different tracers in a universe where structures form hierarchically .
our results reaffirm the validity of cosmic homogeneity on large scales irrespective of the tracers used and strengthens the foundations of the standard model of cosmology .
methods : data analysis - galaxies : statistics - large - scale structure of universe |
historically , attempts at interpreting the observed variability behavior of the broad emission lines in active galactic nuclei ( agn ) have tended to regard the broad emission line region ( blr ) as a stationary entity ( non - evolving ) . by this
we mean that any variability in the observed emission - line intensities has been attributed solely to reverberation ( light - travel - time ) effects within a spatially extended blr .
thus , variations in the measured blr sizes and the amplitude of the line response have been assumed to arise primarily through differences in the temporal variability of the ionizing continuum , which manifests as a difference in shape and fwhm of the continuum autocorrelation function ( acf ) from one event to the next ( e.g. , prez , robinson , & de la fuenta 1992 ) .
in essence , higher frequency continuum variations are better able to probe smaller blr sizes , while larger blr sizes are better probed by lower frequency continuum variations . while a difference in the continuum variability time scale ( e.g.
, the width of the continuum acf ) explains in part some of the observed variability seen in the broad emission lines in agns , it is by no means the complete picture .
indeed , more sophisticated photoionization calculations ( e.g. , see krolik et al .
1991 ; obrien et al .
1995 ; bottorff et al . 1997 ;
kaspi & netzer 1999 ; horne , korista , & goad 2003 ) have modeled the line variability by including the local emission - line response to changes in the overall luminosity and/or shape of the ionizing continuum .
the models presented by carroll ( 1985 ) and netzer ( 1991 ) anticipated the non - linear response of some of the emission lines . in this paper
we illustrate , using a general description of the blr , how large changes in the mean continuum level may explain much of the observed variability behavior of the optical recombination lines . in @xmath4
2 we describe the grid of photoionization models used in this study . in @xmath4
3 we describe the physical origin of line responsivity and its impact on the following issues related to blr variability : the origin and slope of the intrinsic baldwin effect ( @xmath4 3.2 ; e.g. , kinney , rivolo , & koratkar 1990 ) ; variations in the measured continuum emission - line time - delays ( @xmath4 3.3 ; e.g. , peterson et al .
2002 ) , variations in the emission line flux ratios ( @xmath4 3.4 ; e.g. , balmer decrement : tran , osterbrock , & martel 1992 ; h@xmath2/ 4686 : peterson & ferland 1986 ) ; changes in profile shape ( @xmath4 3.5 ; e.g. , wanders & peterson 1996 ) ; and velocity - dependent line intensity ratios ( @xmath4 3.6 ; e.g. , stirpe , de bruyn , & van groningen 1988 ) .
we emphasize that the variations in these quantities discussed here are _ not _ a consequence of reverberation effects within a finite - sized blr , although they are intimately related .
rather , they are due to global changes in the mean ionization state of the blr gas . in @xmath4 4
we discuss the impact of these findings on the purported contribution of a very broad optically thin component to the emission - line response , and other related issues .
it is our hope that many of these observed emission - line variations and trends reported in the literature may be examined in a new light .
we summarize our main findings in @xmath4 5 .
for the purposes of demonstration , we adopted the grid of photoionization computations presented in korista & goad ( 2000 ; hereafter kg00 ) that they used to model the strong uv broad emission lines of ngc 5548 .
the reader is directed there for details ; here we mention some of the salient features of their grid and adopted model . using ferland s spectral synthesis code , cloudy ( v90.04 ;
ferland 1997 ; ferland et al .
1998 ) , kg00 generated a grid of 3249 photoionization models of broad line - emitting entities , here assumed to be simple , single total hydrogen column density slabs ( @xmath5 ) , each of which has constant gas density ( for simplicity ) and a clear view to the source of ionizing photons .
the grid dimensions spanned 7 orders of magnitude in total hydrogen gas number density , @xmath6 , and hydrogen - ionizing photon flux , @xmath7 , and stepped in 0.125 decade intervals in each dimension .
we refer to the plane defined by these two parameters as the density - flux plane .
the choice of spectral energy distribution ( sed ) is described in kg00 ( see also figure a1 in horne , korista , & goad 2003 ) . in the appendix
we discuss the sensitivity of the results presented here to these and other assumptions . in the top four panels of figure 1 we show the equivalent width ( ew ) distributions of the prominent optical broad emission lines of h@xmath3 @xmath06563 , h@xmath2 @xmath04861 , @xmath05876 , and @xmath04686 , namely , those of primary interest in the present paper .
these ews are all referenced to the same incident continuum flux at 1215 and _ are a measure of the efficiency by which ionizing continuum photons are converted into line photons_. the peak in the ew distribution for each line is denoted by a triangle .
note that the ews of the balmer and lines increase with increasing density , especially at the lower values of the photon flux , illustrating the effects of collisional excitation from excited states ( ferland & netzer 1979 ) . in every case ,
the line ews decline in the direction of increasing ionizing photon flux for values greater than those marking the peak ews : @xmath8 , 18.250 , 19.125 , and 20.750 for h@xmath3 , h@xmath2 , , and , respectively .
the rapid decline in the line ews of the two balmer lines for @xmath9 is due to the hydrogen in the fixed
column density slabs becoming fully ionized , a subject we discuss later .
the two helium lines do likewise at slightly larger values of @xmath10 .
the upper left corners of these diagrams represent gas near the compton temperature .
detailed discussions of these and other ew contour plots in the density - flux plane can be found in korista et al.(1997 ) . in order to aid in comparison and to serve as a fixed point of reference ,
the star symbol in each of the panels shows the location of the standard blr parameters from davidson & netzer ( 1979 ) .
the ew distributions of two other hydrogen recombination lines , h@xmath11 @xmath04340 and ly@xmath3 @xmath01216 and minor contributions from ] @xmath01218 and @xmath01216 .
] , are plotted in the bottom panels of figure 1 .
we refer to them from time to time for purposes of comparison . to model the mean uv spectrum of ngc 5548 from the 1993 _ hst _ campaign ( korista et al .
1995 ; hereafter k95 ) kg00 took the `` locally optimally - emitting clouds '' ( loc ) approach of baldwin et al .
( 1995 ) by summing the emission from a weighted distribution of `` clouds '' along the two dimensions of gas density and radius ; kg00 assumed constant column density clouds of @xmath12 @xmath13 .
the cloud distribution functions in gas density and radius are described in kg00 . in brief , the model - integrated spectra included clouds spherically distributed around the continuum source whose gas densities cm . ] and distances from the central source of ionizing photons spanned @xmath14 and 1140 lt - days ; very highly ionized clouds ( @xmath15 ) were excluded from the integration ( see kg00 for details ) .
asymmetric geometric effects , such as emission - line anisotropy ( ferland et al.1992 ; obrien et al .
1994 ) , were not considered .
kg00 showed that this simple model with widely distributed cloud properties was also generally successful in reproducing the observed uv broad emission - line variability . with the choice of cosmological parameters and
adopted galactic reddening along the line of sight to ngc 5548 from kg00 , the mean value of @xmath16 continuum flux from the 1993 _ hst _ campaign ( k95 ) corresponds to @xmath17 @xmath18 .
using the central continuum sed adopted by kg00 , the corresponding mean hydrogen - ionizing luminosity from the 1993 _ hst _ campaign is @xmath19 . at this luminosity
a hydrogen - ionizing photon flux @xmath20 corresponds to a distance from the continuum source of @xmath21 lt - days , and the model blr spans @xmath22 to @xmath23 in this mean continuum state . in this paper
we consider two historical extrema in the @xmath16 continuum flux of ngc 5548 about the above mean level . in 1992 ,
the continuum went into a very low state , @xmath24 , while a high state of @xmath25 was recorded during the 1989 campaign .
this is a factor of about 8.2 change in the continuum flux at 1350 , and we assumed that the sed did not change shape .
while we do present results based on a simple loc model of the blr in ngc 5548 , they should be representative of most general classes of models describing the broad emission lines .
the major trends should be largely model - independent , while the details should be diagnostic of the physical conditions within individual blrs .
summing the line emission along the cloud gas density distribution at each radius , we can derive an effective cloud surface flux of the emission lines at that radius ( see kg00 ; baldwin et al .
figure 2 shows the surface flux of h@xmath2 as a function of radius from the central continuum source for each of the two continuum states .
the slope of cloud emission - line surface flux plotted against the ionizing flux ( @xmath26 ) is a measure of the effective radial responsivity of the emitting gas for small amplitude ( or equivalently , short time scale ) changes in the ionizing continuum ( see goad et al .
similarly , the difference in the two surface flux curves plotted in figure 2 ( in @xmath27 space ) relative to the logarithm of the factor of 8.2 in the incident continuum flux measures a `` global '' response to long - term , large amplitude , changes in the continuum state .
we define this responsivity parameter as @xmath28 this definition of line responsivity converges to the definition given in goad et al.(1993 ) for very small variations in the continuum flux .
note that the upper curve in figure 2 is essentially the lower curve shifted along the x - axis by @xmath29 decades .
a logarithmic slope of @xmath30 in figure 2 corresponds to @xmath31 , whereas flatter slopes correspond to @xmath32 . when integrated over the full line - emitting region , what we call the effective responsivity , @xmath33 , is the logarithmic slope in the relation @xmath34 .
the solid curves in figure 3 illustrate the global responsivities as functions of radius of h@xmath2 ( _ bottom left _ ) , h@xmath3 ( _ top left _ ) , @xmath05876 ( _ top right _ ) , and @xmath04686 ( _ bottom right _ ) for this large change in continuum state .
we emphasize that the major changes in responsivity with radius are not a result of our choice of fixed - column density clouds ; this is discussed below .
the dashed and dotted curves represent the responsivities for small changes in the continuum flux about the high and low continuum states , respectively , and these are discussed in @xmath4 3.1 and @xmath4 3.3 .
it is clear from the global responsivities that larger relative variations of these emission lines surface fluxes due to changes in the continuum state can be expected to occur at larger radii where the incident continuum fluxes are smaller .
we can understand the general trends in line responsivity , with a minimum number of model dependent assumptions , through a study of figure 1 .
we can relate the responsivity to the gradient in the ew contours in figure 1 as @xmath35 the responsivity is proportional to the efficiency in converting a _ change _ in the number of ionizing continuum photons into escaping line photons , and so is in proportion to the change in ew contours shown in figure 1 relative to the incident photon flux . for a fixed scale in cloud distance from the central ionizing source ,
the ew contours in figure 1 shift vertically upward with a decrease in the continuum luminosity , and shift vertically downward with an increase in the continuum luminosity .
clouds in the density - flux plane for which the line ew distribution is either flat or changes little with the incident photon flux have responsivities that lie near 1 ( i.e. , roughly 1:1 line : continuum variations ) for that line ( see , especially , @xmath04686 ) . those clouds whose line ews increase with increasing incident photon flux have responsivities exceeding 1 for that line ( bottoms of the panels for each of these lines except ly@xmath3 ) .
those clouds whose line ews decrease with increasing incident photon flux have responsivities less than 1 for that line : the steeper the gradient , the lower the responsivity .
finally , where the line ews decline very rapidly with incident photon flux in the density - flux plane ( near the central diagonal of the figures ) , the emission line responsivities become negative . to illustrate these effects ,
in figure 4 we superpose a responsivity grey - scale on the ew contours for each of the lines in figure 1 .
the grey - scale representations are : @xmath36 ( _ white _ ) , @xmath37 ( _ light gray _ ) , @xmath38 ( _ medium gray _ ) , @xmath39 ( _ medium dark gray _ ) , @xmath40 ( _ dark gray _ ) , and @xmath41 ( _ black _ , appears only in ) . given that we generally measure @xmath42 for the broad emission lines , this diagram shows how the emissivity and responsivity together constrain the cloud properties , especially their distribution in radius from the continuum source .
we now briefly describe how the responsivities of the optical recombination lines arise . at sufficiently high gas densities and ionizing photon fluxes , each of these four emission lines ews diminish with increasing value of the ionizing photon flux ( fig . 1 ) .
this indicates a lessening efficiency of the conversion of ionizing continuum photons to escaping line photons for those clouds lying nearer to the continuum source .
this is primarily due to the increasing populations and thus optical depths in the excited states of hydrogen and helium with increasing continuum flux ( netzer 1975 ; netzer 1978 ; ferland & netzer 1979 ; ferland , netzer , & shields 1979 ; kwan 1984 ; rees , netzer , & ferland 1989 ; ferland et al .
1992 ; shields & ferland 1993 ) , i.e. , thermalization at large optical depth . at a given continuum flux ,
the optical depth of h@xmath3 is largest , followed by h@xmath2 , @xmath05876 , and @xmath04686 ( rees et al .
various background opacities also play a role ( shields & ferland 1993 ) , and photoionization out of excited states of hydrogen becomes effective for densities exceeding @xmath43 @xmath44 .
in fact , those high density clouds lying above roughly the ew(h@xmath45 contour ( 0.7 dex ) in figure 1 lack a hydrogen ionization front for this reason . as a consequence of these optical depth effects ,
the more substantial relative flux variations in these four emission lines tend to be weighted by clouds lying at larger radii
. this latter point should be true to some extent for most emission lines , since they too suffer from diminished efficiency in the conversion of ionizing continuum photons into escaping line photons at smaller radii where the gas densities of clouds that emit the lines must be higher .
refer to figure 2 of kg00 for the ew contours in the density - flux plane for the stronger uv emission lines .
the @xmath02800 emission line is most like the balmer lines in this respect , and we would expect that line to have a similarly low responsivity ( see figure 2b in kg00 ) .
this is indeed generally observed ( e.g. , see clavel et al .
clouds fully ionized in hydrogen ( such as those lying near or within the rapidly declining ew contours running diagonally in the density - flux planes in figure 1 ) , with their very low ( @xmath46 0 ) or even negative responsivities , might in principle be arranged in such a way as to substantially reduce the line responsivity at the larger radii .
however , these low efficiency clouds will not contribute much light and therefore will add little to the integrated line response unless their covering fractions are very large . for time - scales of the order of the light - travel time across the blr , the power spectrum of the continuum fluctuations coupled with light - travel time effects across a finite geometry will also factor into the variations of an emission line s flux and perhaps profile . however , _ it is the responsivity that establishes the outer envelope of these variations_. light travel time effects can only act to reduce or dilute the observed line response .
we reiterate that the variations in the line intensity ratios and line profiles presented here are _ not _ due to light travel time effects ( e.g. , see robinson , prez , & binette 1990 ; prez , robinson , & de la fuente 1992 ; pogge & peterson 1992 ) or other geometrical effects ( e.g. , line anisotropy , obrien , goad , & gondhalekar 1994 ) , but are rather due to differences in the time - averaged photon flux incident on a spatially extended blr and their effect on the line emissivity .
krolik et al .
( 1991 ) , pogge & peterson ( 1992 ) , and gilbert & peterson ( 2003 ) have found that reverberation effects on the @xmath47 relation can be largely removed by correcting the emission - line light - curves for their mean time - delays .
if this is indeed the case , then any residual line continuum flux correlations must be primarily due to the responsivity of the gas in the blr , as we have defined here .
however , if a line s transfer function has a long tail in time delay ( in comparison to the important continuum variability time scales ) , simply removing the lag may not sufficiently correct for this geometric dilution of the line response . the observed lag - corrected emission - line response will then be a function of the line s responsivity , as defined here , and the prior continuum history .
more advanced methods , such as that introduced by horne et al .
( 2003 ) , may then be needed to disentangle the effects .
finally , it is apparent from figs . 1 , 3 , and 4 that @xmath04686 should be the most responsive emission line of the four optical emission lines discussed here , followed by , h@xmath2 , and lastly h@xmath3 .
this is observed ( e.g. , see wamsteker et al .
1990 ; dietrich et al . 1993 ; kollatschny 2003 ) .
the model effective global responsivities of these four emission lines are listed in column ( 2 ) of table 1 . to illustrate the trends within the balmer line series
, we also plot the ew distribution of h@xmath11 @xmath04340 within the density - flux plane in figure 1 , list its effective responsivity in table 1 , and illustrate its responsivity in the density - flux plane in figure 4 .
ly@xmath3 @xmath01216 is also a recombination line , but with different physics , and so we provide the same information for it .
figure 3 shows that the responsivity of h@xmath2 is expected to be a strong function of radius , varying from @xmath46 0.2 in the inner blr to @xmath46 1 at the outer radius .
more generally , this is indicated by the gradient in the ew contours with respect to the ionizing photon flux in figure 1 ( see also eq .
2 ) , and coarsely illustrated in figure 4 .
another interesting point raised by figure 3 is that the h@xmath2 line response should be significantly stronger during low continuum flux states than during high ones for similar continuum fluctuation amplitudes .
this is illustrated in figure 3 by the dotted and dashed curves representing the low and high continuum state h@xmath2 line responsivity , respectively , computed for small ( 0.1 dex ) variations about the high and low states .
the effective responsivities in the high and low states are given in columns ( 3 ) and ( 4 ) of table 1 .
k. horne ( 2002 , private communication ) and cackett & horne ( 2005 ) recovered the luminosity - dependent delay map for the 13 yr monitoring campaign of the h@xmath2 broad emission line of ngc 5548 , and found a similar relation between the amplitude of the line response with continuum state .
figure 3 and columns ( 3 ) and ( 4 ) of table 1 shows that all of these lines display generally larger effective responsivities at larger distances ( lower incident photon fluxes ) and during lower continuum states .
additionally , the amplitude of the change in effective responsivity with continuum state differs substantially between the balmer and helium lines , and this should be diagnostic of the physical conditions within the blr .
by contrast , ly@xmath3 shows very little variation in responsivity with cloud parameters ( fig .
4 ) , and as a consequence shows a very modest change in @xmath33 with continuum state ( table 1 ) .
we elucidate the change in the radial responsivity with continuum state shown in figure 3 , using @xmath04686 by way of example . in both continuum states
, the responsivity at large radii is nearly independent of radius and continuum state .
this is simply a reflection of the general lack of dependence of the line ew with incident continuum flux for @xmath48 ( see figure 4 ) .
this is in contrast to the case of the balmer lines , especially h@xmath3 .
note also in figure 4 that most of the variations of @xmath49 ( ) with incident photon flux occur for @xmath50 ; here in particular @xmath49 declines with increasing @xmath51 .
this dependence of the responsivity on incident continuum flux makes this line s radial responsivity at small distances quite dependent on the continuum state as well .
these effects are also illustrated in figure 3 .
the `` knee '' in the radial responsivity of occurs at the radius corresponding to @xmath52 for each continuum state , or @xmath53 ( 7 lt - days ) in the mean continuum state .
simply put , strong anti - correlations between the line ew and the incident photon flux will correspond to anti - correlations between @xmath49 and the incident photon flux , which in turn will result in similar anti - correlations between @xmath49 and the continuum state .
likewise , @xmath33 will anti - correlate with the continuum state , as weighted by the line s radial luminosity function . based on single power - law cloud distribution models , obrien ,
goad , & gondhalekar ( 1995 ) also found `` a decrease in the mean response when the continuum flux increases . '' by definition
, a time - varying responsivity @xmath54 implies that the line transfer function , @xmath55 ) , is time - dependent ( i.e. , non - stationary ) . in this case
the time dependence is due to changes in the time - averaged photon flux incident on a spatially extended blr .
we therefore expect the measured lags of the emission lines and their profile shapes to depend on the luminosity state of the continuum ( @xmath4 3.3 & @xmath4 3.4 ) . we note that unless the velocity field is self - similar in radius ( goad et al .
1999 ) , a profile shape that depends on the continuum luminosity state fails a hysteresis test ( see wanders 1994 ; perry , van groningen , & wanders 1994 ) .
this is true without invoking a redistribution of material within the blr or other geometrical effects as suggested by wanders & peterson ( 1996 ) .
detailed analyses of the agn broad emission - line response to continuum variations in spectrophotometric monitoring campaigns show that in all cases the responsivity in most measured lines is @xmath56 , even after accounting for light travel - time effects ( e.g. , krolik et al.1991 ) . referring to eq . 2
, we see that this requires the line equivalent width to diminish with increasing continuum flux and vice versa : the intrinsic baldwin effect ( @xmath57 ; see kinney , rivolo , & koratkar 1990 ; krolik et al .
1991 ; pogge & peterson 1992 ; peterson 1997 ; gilbert & peterson 2003 ) . in concurrence with gilbert & peterson the present work suggests that the origin of the intrinsic baldwin effect is due either in whole or in large part to the line s responsivity being less than 1 throughout the majority of a distribution of largely ionization - bounded clouds ( as shown in figure 1 ) , i.e. , without the _ necessity _ of a large population of matter - bounded clouds ( shields , ferland , & peterson 1995 ) , although we do not rule these out .
as discussed above we expect @xmath58 for most lines , and thus most lines should demonstrate an intrinsic baldwin effect , since the intrinsic baldwin effect index @xmath2 is related to the responsivity as @xmath59 .
based on single power law cloud distribution models , obrien , goad , & gondhalekar ( 1995 ) showed that an intrinsic baldwin effect can be induced in any line by changing its responsivity with continuum level . while we concur with these findings , we find a more general and significant origin for this effect , namely @xmath58 as defined and described in @xmath4 3 .
our simple model for the blr of ngc 5548 predicts an @xmath33 for h@xmath2 of 0.54 in the 1989 high state and 0.77 in the 1992 low state ( columns ( 2 ) ( 4 ) , table 1 ) .
that is , we expect @xmath33 and so the slope in the intrinsic baldwin effect to be a function of continuum state .
compare the dotted and dashed lines in figure 3 .
this predicted range in @xmath33 approximately brackets that measured in the same object by gilbert & peterson ( 2003 ) of @xmath60 .
a line s equivalent width is usually measured with respect to the local continuum , and gilbert & peterson ( 2003 ) found that the underlying continuum becomes bluer when brighter in this object , such that @xmath61 .
they then suggested that since the @xmath01350 continuum lies closer to the ionizing continuum , it might be better to reference the h@xmath2 flux to it , and found the relation @xmath62 .
our simulations assumed that the sed remains constant as its amplitude changes .
thus , while it is difficult to make detailed comparisons , we expect that sed changes are a secondary effect on the line responsivity ; we discuss the impact of a variable sed on line responsivity in @xmath4 4.2 .
the point remains that line responsivity , as defined here , is likely to be responsible for much of the observed intrinsic baldwin effect .
what can the intrinsic baldwin effect tell us about the blr ?
because the ew contours of the balmer lines in particular in figure 1 are so nearly proportional to the incident continuum flux , we suggest that measurements of the responsivities of these emission lines will point to the distribution of the optical recombination line - emitting entities in the density - flux plane . low density gas ( @xmath63 @xmath44 ) illuminated by low values of the ionizing photon flux ( @xmath64 @xmath13s@xmath65 ) , and gas at any density illuminated by even lower values of @xmath51 ( @xmath66 ) will emit these optical recombination lines , but they _ will not respond appropriately _ to variations in the ionizing continuum .
the responsivities of these clouds in these lines are @xmath46 1 and @xmath67 , respectively .
such models would fall far short of reproducing the observed variations in h@xmath2 , _ independent _ of any considerations of the line luminosity or lag .
next , the observed mean responsivity of the optical @xmath04686 line , @xmath68 ( see , for example , dietrich et al .
1993 ; also krolik et al.1991 ) , necessarily implies that a significant fraction of this line forms in gas exposed to high incident ionizing continuum fluxes on average ( @xmath69 @xmath13 s@xmath65 , or @xmath70 light days in ngc 5548 ) , as well as relatively high densities .
this is just as its measured lag indicates ( clavel et al . 1991 ; k95 ) .
thus , figures 3 and 4 predict a positive correlation between the lag of each of these lines and their responsivity , @xmath33 , for a particular continuum state ( see also carroll 1985 ) .
so together the luminosity , lag , _ and _ responsivity ( intrinsic baldwin effect ) of each of the broad emission lines for a range of continuum states should constrain the distribution of blr cloud properties in gas density , distance from the ionizing continuum source , and column density .
the quasar tomography method of horne et al .
( 2003 ) is particularly well suited to this task , and we discuss this further in @xmath4 4.3 .
netzer & maoz ( 1990 ) reported that different lags for the broad emission lines of ly@xmath3 and were found for each of the three continuum variability events during the 1989 campaign of ngc 5548 ( clavel et al .
1991 ) . based on data from an earlier optical campaign ( netzer et al .
1990 ) , netzer & maoz also reported a significantly different lag for h@xmath2 than did peterson et al .
while acknowledging that a variable continuum acf will change the lag , netzer & maoz ( 1990 ) suggested that at least part of the origin of these time - dependent lags lay in a non - linear emission - line response . over the full 13 yr monitoring campaign of ngc 5548 peterson et al .
( 2002 ; see also peterson et al.1999 ) found that the h@xmath2 emission - line lag scales with the mean continuum level , such that @xmath71 , albeit with much scatter .
this positively correlated behavior of the lag with continuum state has been called `` breathing . '' after positing that the uv continuum is more intimately connected to the driving ionizing continuum and accounting for the relation that they found between the @xmath05100 and @xmath01350 continuum bands ( @xmath72 ) , the relation @xmath73 resulted .
peterson et al.(2002 ) then proposed that @xmath74 might be a natural consequence of a simple ionization parameter - type argument that @xmath75 .
cackett & horne ( 2005 ) , too , have found evidence of breathing in their transfer function analysis of the 13 yr spectroscopic dataset of peterson et al .
a breathing blr has also been proposed by goad et al .
( 1999 ) to explain line profile variation differences between and in ngc 3516 . in this subsection
, we address a likely origin of breathing blrs and comment on the conclusions drawn by peterson et al .
( 2002 ) in regards to this issue .
the single power law cloud distribution models of obrien , goad , & gondhalekar ( 1995 ) predicted a relation between line lag and incident continuum level ( for a fixed sed ) due to a non - linear emission - line response .
the results presented here for the optical recombination lines based on more general models arrive at the same conclusion .
the shift in the lag as a function of the continuum state is a reflection of an outward shift in the line s luminosity - weighted radius to gas at larger radii that is better able to reprocess the increased continuum luminosity into line emission , and vice versa in low continuum states .
the ew contours in figs . 1 and 4 shift vertically as the source luminosity changes , causing corresponding changes in the radial responsivity and hence responsivity - weighted radii .
we therefore expect that elevated continuum states will be accompanied by an increase in the line lag and a decrease in the line responsivity .
breathing will not occur for lines for which the responsivity as a function of radius and ionizing continuum level is constant .
a constant @xmath76 implies that the emissivity - weighted and responsivity - weighted radii are the same ( see goad et al .
1993 ) , and therefore the measured lag is independent of the continuum state
. therefore , if after correcting for reverberation effects , the measured lag varies with luminosity , then so must its effective responsivity .
although gilbert & peterson ( 2003 ) did not note any evidence for changes in @xmath33 with continuum level , b. peterson ( 2003 , private communication ) finds that a detailed analysis of the 13 yr campaign on a year - to - year basis indicates changes in the line continuum flux slope .
it is also significant that while blrs whose lines breathe must show an intrinsic baldwin effect for those lines , the converse is not true .
for example , for the case of @xmath77 everywhere , no breathing will occur , but an intrinsic baldwin effect will be present . the two triangles in each panel of figure 3 mark the positions of the responsivity - weighted radii ( @xmath78 ) of h@xmath2 and the other three emission lines for the two continuum states .
in the case of h@xmath2 , it is larger in the high continuum state by a factor of @xmath46 1.6 relative to the low state .
this is about the same ratio of luminosity dependent lags observed in this line in ngc 5548 for the same two continuum states ( @xmath46 1.7 ; peterson et al .
2002 ; see also wanders & peterson 1996 ; cackett & horne 2005 ) , although we note that the measured lag is expected to be smaller than @xmath79 by a factor of a few ( e.g. , see prez et al .
1992 ) .
the effective line responsivities for the high and low continuum state are given in columns ( 3 ) and ( 4 ) of table 1 .
as discussed in @xmath43.1 , the low - state responsivity is higher in every case , and figure 3 shows that the line s radial responsivity and @xmath78 shift to larger distances in higher continuum states . for this particular model , the responsivity - weighted radius scales as @xmath80 , with @xmath81 0.15 , 0.23 , 0.38 , and 0.44 for h@xmath3 , h@xmath2 , , and , respectively .
these relations should be diagnostic of the physical conditions within the blr .
the general relation proposed by peterson et al .
( 2002 ) , above ( @xmath82 ) , is expected to hold only if the line responsivity is constant over a large portion of the blr with a steep break to lower responsivity ( see for example , in figure 3 ) , and so is not the general case .
the breathing amplitude ( e.g. , the value of @xmath11 ) is expected to depend on the responsivity and the extent to which the responsivity changes with both radius and continuum state in the vicinity of @xmath78 ( compare dotted and dashed curves for each of the lines in figure 3 ) . using the relation between the @xmath05100 and @xmath01350 continuum fluxes for ngc 5548 ( gilbert & peterson 2003 ; but see also peterson et al . for a slightly different value ) , and assuming that @xmath83 is a proxy for the driving continuum , k. horne ( 2002 , private communication ) and cackett & horne ( 2005 ) find a similarly weaker relation between the h@xmath2 lag and continuum luminosity state for the 13 yr campaign of ngc 5548 as do we .
in addition , since @xmath11 is related to the responsivity , it too changes with continuum level ( see also obrien et al . 1995 ) .
finally , we note that the expected larger line response during low continuum states ( @xmath4 3.1 ) will be enhanced by the fact that an effectively smaller line - emitting region , due to this breathing , respond more coherently to a given continuum variation .
the opposite hold for variations about high continuum states .
the combined effect can be investigated in data sets such as the 13 yr h@xmath2 light curve presented in peterson et al .
( 2002 ; their figure 1 ) : more coherent and larger amplitude responses in h@xmath2 should generally be found at low continuum states , while just the opposite should result during high continuum states .
one might also find differences in line responsivity between two comparably low continuum states that immediately straddle a high continuum state in time , due to light travel time effects that are not removed from the second low state by simply correcting for the lag . in this section , we investigate the expected changes in the integrated line flux ratios due to changes in the continuum state .
it should now be apparent that such line ratio variations will result simply because the responsivities differ between the emission lines .
lines with greater responsivities vary with greater amplitudes with changing continuum state , and so line flux ratios can be expected to change as well .
other factors may play a role , but responsivity effects will be present regardless .
figure 5 illustrates the h@xmath3/h@xmath2 line flux ratio varying across the density - flux plane , from roughly 17 ( near the triangle at the bottom ) to approximately 1 ( near coordinates : [ 13.5 , 22.5 ] ) .
the classical nebular theory of hydrogen balmer lines ( menzel case b ) does not apply to the gas occupying the blr .
for example , at coordinate [ 12.0 , 19.5 ] in the density - flux plane ( corresponding to a distance of 22 lt - days in the mean continuum state ) , case b predicts h@xmath3 and h@xmath2 fluxes that are 20.8@xmath84 and 30.4@xmath84 larger than that emitted by the cloud . the case b h@xmath3/h@xmath2 flux ratio is 2.35 , whereas the ratio predicted by cloudy is 3.44 . over a significant portion of the density - flux plane _
this ratio is inversely proportional to the incident continuum flux _
, as expected from the optical depth effects mentioned earlier , and in agreement with the findings of rees , netzer , & ferland ( 1989 ; see also netzer 1975 ) . this being the case , it is easy to imagine this ratio changing in response to variations in the incident continuum luminosity .
thus , differences in responsivity among the balmer lines , as illustrated in figs .
1 , 3 , 4 and as listed in table 1 , probably account for most of the observed steepening in the broad - line balmer decrement as the continuum drops into low states , and flattening in high states .
there have been many such broad line balmer decrement / continuum state correlations reported in the spectra of variable seyfert 1 galaxies ( see , for example , antonucci & cohen 1983 ; ferland , korista , & peterson 1990 ; wamsteker et al .
1990 ; tran , osterbrock , & martel 1992 ) . integrated over this model s full broad line - emitting region , the h@xmath3/h@xmath2 flux ratio changes from 4.9 in the low state to 3.7 in the high state .
while the particulars are model dependent , the trend generally is not .
the balmer decrement s variations with the incident continuum level should be sensitive to the distribution of clouds in the density - flux plane .
furthermore , we suggest that a seyfert 1 galaxy s sub - classification as seyfert 1.0 , 1.5 , 1.8 , or 1.9 ( as defined in osterbrock 1989 , and references therein ) may at least in part be due to the relative flux of ionizing photons present within their broad - line regions at the time of observation . that is , any given spectrum of a seyfert type 1 agn may pass through some of these sub - classifications as its continuum luminosity fluctuates from high ( seyfert 1 or 1.5 ) to low continuum states ( seyfert 1.8 or 1.9 ) .
in high continuum luminosity states , a larger broad - line contribution to the total ( broad + narrow ) balmer line profile will be present and the broad balmer line decrement will be flatter . in low continuum luminosity states , the broad - line contribution to the total balmer line profile will be relatively much smaller , and the broad balmer line decrement will be steeper .
ngc 5548 is generally classified as seyfert 1.5 .
however , the order - of - magnitude continuum fluctuations that occurred over the 13 yr optical monitoring campaign ( peterson et al .
2002 ) probably drove the balmer emission - line spectrum from seyfert type 1.8 in 1992 to perhaps seyfert type 1.0 in 1998 .
as described above , the balmer lines should show less flux variation with continuum state than , which in turn should show less variation than .
this trend is generally observed ( peterson & ferland 1986 ; ferland , korista , & peterson 1990 ; wamsteker et al .
1990 ; dietrich et al . 1993 ; peterson et al . 1993 ; kollatschny 2003 ) .
particularly interesting is that observed changes in the h@xmath2/ flux ratio may , at least in part , due to _ differences in the way these two lines respond to the level of the ionizing continuum flux _ , rather than to changes in the shape of the ionizing sed .
for the model presented here , this ratio drops by nearly a factor of 2 from the low to high continuum states . to a lesser extent , this is likewise true for the / line ratio , which changes by a factor of 1.3 .
such a responsivity effect may have played a role in the observation of a large increase in the @xmath04686 equivalent width in comparison to h@xmath2 during a large increase in the continuum luminosity in ngc 5548 ( peterson & ferland 1986 ) and likewise in a similar event reported by peterson et al .
( 2000 ) in ngc 4051 .
while the broad line /h@xmath2 flux ratio is sensitive to the ionizing sed , the usual photon - counting ( zanstra ) argument used to measure the 13.6 to 54.4 ev continuum sed applies strictly when these lines are emitted in gas whose properties allow for responsivities @xmath85 , as would be true if case b applied for both lines .
however , these conditions _ do not _ apply for the balmer lines emitted within the blr , but they can be found within the narrow line regions of agns ( see the ew distributions of the recombination lines in ferguson et al .
1997 ) or within galactic nebulae . on the other hand , the emissivity does not stray too far from that expected from case b for most conditions expected in the blr , as indicated by its comparatively featureless ew distribution in the density - flux plane for @xmath86 ( see also bottorff et al .
2002 ) .
finally , as a matter of general interest we note that the ly@xmath3/h@xmath2 flux ratio should correlate with the continuum flux , and apparently this is the case in the well - studied agn ngc 5548 ( wamsteker et al .
the ew contours of ly@xmath3 in the density - flux plane ( see figure 1 , _ bottom right _ ) do not show as strong continuum flux dependencies as do the balmer lines , and so @xmath49(ly@xmath87(h@xmath2 ) can be expected , with larger differences to come from gas nearer to the continuum source where the responsivity of h@xmath2 is small ( see also figure 4 ) . the combined conclusions of pogge & peterson ( 1992 ) and gilbert & peterson ( 2003 ) would indicate an observed relation of @xmath88ly@xmath89h@xmath90 in ngc 5548 .
our simple model ( see table 1 ) predicts a similar relation @xmath88ly@xmath89h@xmath91 , which is steeper in high continuum states and shallower in low ones .
shields & ferland ( 1993 ) discussed some of the mechanisms behind the destruction of ly@xmath3 emitted by broad - line clouds .
pogge & peterson also found that in fairall 9 @xmath33(ly@xmath92 for a factor of @xmath4610 change in the observed @xmath01338 continuum , similar to the value reported here ( 0.74 , table 1 ) for our simple model .
figure 4 shows that @xmath49(ly@xmath3 ) is nearly independent of gas density and incident photon flux over a wide range in these parameters .
while the dynamics of the gas emitting the broad emission lines are not yet understood , the importance of a virial component seems to be reasonably well established ( peterson & wandel 2000 ) . here
, we have adopted the simple relationship @xmath93 assuming randomly inclined circular orbits and a central mass of @xmath94 solar masses , chosen to approximately reproduce the observed broad - line fwhm of h@xmath2 in ngc 5548 .
this choice of dynamical model is motivated by simplicity for purposes of demonstration .
the low - state continuum emission - line profiles of the four lines under study , as predicted by our simple loc model line emissivity and velocity field , are shown in figure 6 , normalized to their peak intensities .
we note that the flat tops to these profiles are artifacts of the assumption of circular orbits , and their extents are determined by the local keplerian velocity at the distance of the model s outer radius ( 140 lt - days ) given the assumed central mass .
the ew distributions shown in figure 1 , and most notably the continuum flux at the onset of significant thermalization , are roughly reflected in the widths of the emission - line profiles in figure 6 , as well as in their observed lags ( dietrich et al .
1993 ; peterson & wandel 1999 , 2000 ; fromerth & melia 2000 ) . is the broadest line ( with the shortest lag ) , followed by , then h@xmath2 , and finally h@xmath3 ( with the longest lag ) .
this is as generally observed in broad - line agns ( stirpe 1990 , 1991 ; crenshaw 1986 , shuder 1982 , 1984 ; osterbrock & shuder 1982 ; boroson & green 1992 ; kollatschny 2003 ) .
krolik et al .
( 1991 ) , peterson & wandel ( 1999 , 2000 ) , and onken & peterson ( 2002 ) present highly correlated lag
line width relations that indicate virial - like motions in the vicinity of a supermassive central object .
the fwhm of these four emission lines , as well as h@xmath11 and ly@xmath3 , are listed in columns ( 5 ) and ( 6 ) of table 1 for the high and low continuum states , respectively .
the trend in line widths is in general agreement with those found in the models presented by rees , netzer , & ferland ( 1989 ) . since the lines under study here are recombination lines involving just one or two electrons , they are emissive over very broad areas within the density - flux plane , unlike the collisionally excited metal lines which are very sensitive to the ionization parameter .
the balmer and lines span essentially equal areas of significant emissivity within the density - flux plane _ without regard to ionization parameter _ , and that of does likewise for @xmath95 ( see figure 1 ) . while it is also true that a cloud of a given column density will become fully he@xmath96 at a higher ionization parameter than for h@xmath97 , this difference is small ( see figure 1 ) . by far most of the differences in emissivity among the balmer and lines
are due to the optical depth effects discussed above , and this is also true for the lines except for clouds with @xmath98 . stated another way , if all of these lines were emitted with their case b emissivities within the blr , much smaller differences in their profiles and lags would be expected .
figure 7 shows the long term ( global ) responsivities for each of the four emission lines as a function of radial velocity across their line profiles , for our simple velocity field . because of the larger response in the line cores ,
the broad emission - line component of the balmer lines should become narrower / peakier in higher continuum states , and broader / less peaky during low continuum states ( cf . , h@xmath3 variations in ngc 5548 ; stirpe et al .
1988 ) ; refer also to the model high and low continuum state fwhms as listed in columns ( 5 ) and ( 6 ) in table 1 .
in fact for this particular model , a continuum state that is a factor of 12@xmath84 brighter than the low state produces a fwhm in h@xmath2 that is 1000 @xmath99 narrower than that in the low state . the response across the profile of is _ much higher _ , although again this emission line would be expected to be somewhat narrower during higher continuum states .
the response across the profile of is not as broad as and not as peaky as the balmer lines . in every case , _ we expect the response in the line cores to be stronger than in the line wings_. the relative change in the fwhm of the emission line with continuum state is effectively a reflection of the `` breathing '' blr as caused by changes in the line responsivity , as we discussed in @xmath4 3.3 .
obrien , goad , & gondhalekar ( 1995 ) similarly linked non - linear emission - line response to line profile variability , although this general conclusion was based on single power law cloud models .
since the responsivity of these lines is expected to be radially dependent and this function differs between emission - line species , we expect radial velocity - dependent variations in the line intensity ratios .
these effects are illustrated in figure 8 , which shows the velocity - dependent balmer decrement ( h@xmath3/h@xmath2 ) for the low - state ( _ dotted line _ ) and high - state ( _ solid line _ ) continuum .
note that in both cases the balmer decrement is steeper in the line core than in the line wings , and the balmer decrement shows larger changes with continuum level in the line core ( from 5.6 to 4.0 ) , than in the line wings ( from 3.0 to 2.7 ) .
the first of these effects has been well - established ( see , for example , shuder 1982 , 1984 ; osterbrock & shuder 1982 ; van groningen 1984 , 1987 ; crenshaw 1986 ; stirpe 1990 , 1991 ; kollatschny 2003 ) , and we are aware of just one study of the time - dependent variability of the balmer decrement as a function of radial velocity ( stirpe , de bruyn , & van groningen 1988 ) . both behaviors imply that _ the velocity field must decrease with increasing radius from the central source_. as just presented above , a weaker response in an individual line s wings in comparison to its core ( fig .
7 ) would also argue for such a velocity field .
these findings are consistent with recent reverberation studies that find a relationship between the emission - line lag and rms profile width for multiple lines in individual agns , which implies that the velocity field is dominated by the central massive object ( peterson & wandel 1999 , 2000 ; fromerth & melia 2000 ) .
the results presented here also predict a smaller h@xmath2/ intensity ratio in the line wings than in the core ( fig .
the ew contours in figure 1 coupled with a velocity field that diminishes with increasing distance from the central source provide possible explanations for this generally observed phenomenon ( shuder 1982 , 1984 ; osterbrock & shuder 1982 ; crenshaw 1986 ; van groningen 1984 , 1987 ; rees , netzer , & ferland 1989 ; kollatschny 2003 ) .
these lines ews peak up at similarly high densities , but the line peaks up at significantly higher photon fluxes ( smaller radii ) , because of its lower optical depth .
this is responsible for this line s relatively stronger wings ( see also netzer 1978 ; kwan 1984 ; carroll 1985 ; carroll & kwan 1985 ) .
whether or not the specific dependencies of the continuum state / responsivity effects described in this section are present in the spectra of ngc 5548 or other variable seyfert 1 agns is not particularly important .
we discuss the sensitivity of line responsivity to model assumptions in the appendix .
instead , the emphasis of this paper lies mainly in illuminating the mostly untapped wealth of the potentially powerful physical diagnostics present in high quality spectrophotometric data sets of the optical recombination lines , and that their emissivities sensitivity to the continuum flux makes them particularly effective diagnostics of the radially dependent physical conditions within the blr .
that the blr may contain a significant component of ionized gas that is optically thin to hydrogen - ionizing photons has been suggested by investigators over many years . in nearly every case
there is an association of this gas with that emitting the very broad wings of the lines .
most recently , sulentic et al .
( 2000 ) reported that in the quasar pg 1416@xmath100129 the fwhm of the broad emission - line profile of h@xmath2 increased by 50% from a high to a low continuum state separated in time by 10 yr .
their interpretation of this is that the wings of the balmer lines are dominated by emission from an optically thin `` very broad line region '' ( vblr ) , while the core of the line is dominated by more highly variable optically thick gas .
these are usually thought of as two distinct line - emitting regions .
ferland , korista , & peterson ( 1990 ) , peterson et al .
( 1993 ) , and corbin & smith ( 2000 ) made similar arguments based on balmer line continuum variations , while gondhalekar ( 1987 , 1990 ) , obrien , zheng , & wilson ( 1989 ) , & prez , penston , & moles ( 1989 ) found less variable wings in ly@xmath3 and @xmath01549 of higher redshift quasars .
kassebaum et al .
( 1997 ) found similarly stronger variations in the line core than wings of h@xmath2 in mrk 335 .
it should be noted that while the conclusions regarding the core versus wing variations from monitoring campaign studies such as peterson et al.(1993 ) and kassebaum et al .
( 1997 ) were based on analyses of _ time - resolved _
spectral observations of the continuum and h@xmath2 , most of the above studies relied on very few , often just two , observation epochs .
without sufficient time resolution of the variability , it is difficult to assess the actual line response .
additional arguments for an optically thin line - emitting region have been put forth through the analyses of line flux ratios in the profile wings : optical balmer and lines ( marziani & sulentic 1993 ; corbin 1997 ) , balmer and @xmath01549 ( corbin 1995 ) , balmer and ly@xmath3 line profiles ( zheng 1992 ) , and balmer and @xmath08446 profiles ( morris & ward 1989 ) .
while at any given continuum state the broad emission - line region may well contain clouds fully ionized in hydrogen , as the present model does , the results presented here suggest that there is no great need for a major line - emitting component consisting of vast amounts of optically thin gas .
most of the effects mentioned just above regarding the balmer and helium recombination lines can be explained by the effects we described in @xmath4 3 : weak balmer line emission and response need not originate in clouds optically thin to the lyman continuum . while an emission line such as @xmath0774 is likely to be emitted mainly within clouds that lack a hydrogen ionization front ( e.g. , see hamann et al.1998 ; korista et al .
1997 ) , we have also pointed out that by their very nature fully - ionized clouds are inefficient emitters of hydrogen recombination lines ( refer again to figure 1 ) , and so are unlikely to be important energetically even if their covering fraction is large .
this is especially true in the face of a variable ionizing continuum source ; clouds will more likely be either ionization - bounded or too overionized to emit balmer lines of any significance .
it is also significant that the matter - bounded clouds in figure 4 either have small positive ( _ lightest gray _ ) 0 lie above the diagonal line @xmath101 and also within the lightest gray shaded region . ] or negative ( white ) responsivities .
even if present , the latter type clouds never make their presence known in these emission lines , which is not surprising given their extremely low emitting efficiencies . in regards to the comparisons of balmer and ly@xmath3 line profiles , as mentioned in @xmath4 3.4 , the ew contours of ly@xmath3 in the density - flux plane ( see figure 1 , _ bottom
right _ ) do not show the strong continuum flux dependencies of the balmer lines .
we suspect that this , along with the presence of a radially decreasing velocity field , accounts for the very large ly@xmath3/h@xmath2 ratios found in the emission - line wings by zheng ( 1992 ) , rather than these ratios being due to a major constituent of optically thin emitting gas . finally , we also point out that many of the above studies assumed that case b emissivities apply to the hydrogen lines emitted by optically thick clouds in the blr and attributed significant deviations from expected ratios to an optically thin emitting region .
we conclude that a separate optically thin vblr may be unnecessary , and in any case for the balmer lines is probably energetically unimportant .
the wings of the hydrogen recombination lines likely represent nothing more than the inner blr where the balmer line emission is inefficient and the responsivity low .
paltani & courvoisier ( 1994 ) and paltani & walter ( 1996 ) showed unambiguously that the uv continuum becomes bluer when the flux is higher in variable agns .
this effect has also been found in the sloan digital sky survey quasar sample ( vanden berk et al .
2004 ) . in the particular case of ngc 5548 ,
maoz et al .
( 1993 ) reported that , after accounting for the balmer continuum and uv emission , the 18002400 continuum becomes bluer with increased continuum flux .
additionally , peterson et al . ( 2002 ) and gilbert & peterson ( 2003 ) found that after careful removal of the contaminating non - variable stellar optical continuum flux , the uv - optical continuum in ngc 5548 becomes bluer as it becomes brighter : @xmath102 .
extrapolation of this relation to higher energies might indicate a hardening of the incident ionizing continuum during brighter continuum states .
marshall et al .
( 1997 ) found larger amplitude variations in the extreme ultraviolet ( @xmath46 100 ) than at 1350 during a week - long period of the 1993 ngc 5548 monitoring campaign .
alternatively , as suggested by paltani & walter ( 1996 ) , it may indicate the presence of a constant ( or more slowly varying ) continuum or pseudocontinuum component whose contribution is greater at longer wavelengths .
korista & goad ( 2001 ) found that 20%30% of the observed effect may be due to reverberation of the diffuse continuum emission of the broad line clouds .
in the simulations presented here , the sed of the continuum incident on the broad - line clouds is the same for both high and low continuum states . how an emission line s responsivity would change with an _ incident
_ sed that changed shape as the uv - optical continuum level varied will depend on the details of how the flux of photons important to the creation and destruction of the line varies .
we briefly discuss these effects here .
the shape of the ionizing continuum determines the heating for given values of gas density , ionizing photon flux , and chemical abundances .
all else being equal , harder incident continua result in higher electron temperatures , and so major cooling lines ( e.g. , @xmath01549 ) might be expected to increase their energy output in response .
one should also keep in mind that the incident flux at energies corresponding to the _ balmer _ continuum is important to the destruction of hydrogen emission lines for the high gas densities encountered in the blr ( shields & ferland 1993 ) .
also , harder continua during brighter uv - opt continuum states might also mean comparatively more photons above 1 ryd and/or 4 ryd , favoring the production of hydrogen and helium recombination lines .
kaspi & netzer ( 1999 ) considered an sed whose break near 4 ryd moves to higher energies for brighter uv continuum states ( their figure 8d ) , with the effect of increasing the 4 ryd/1 ryd photon flux ratio with increasing @xmath01350 continuum luminosity .
however , in this variable sed scheme , the flux above 20 ryd ( 270 ev ) remained constant .
a comparison of their figs . 7 ( fixed sed ) and 9 ( variable sed ) that show their model predictions of five uv emission lines light curves versus observations illustrates some of the aforementioned effects of a variable continuum sed on line responsivity . while they included @xmath01640 in their simulations , such was not the case for the balmer lines .
their prescription for a variable continuum sed increased the responsivities of the and ly@xmath3 lines , whereas those of @xmath01549 , ] @xmath01909 , and @xmath02800 decreased , presumably because the variations in the continuum beyond @xmath46 150 ev were smaller in their variable sed scheme than for the constant sed .
it is also apparent that the lags of and increased in the higher continuum states , because of a shift to larger responsivity - weighted radii .
the method of `` quasar tomography , '' proposed by horne et al.(2003 ) , is a unification of reverberation mapping and photoionization physics .
its goal is to place physical constraints on the recovered two - dimensional transfer function map @xmath103 of the broad emission lines , as well as to recover the gas distribution described by a five - dimensional map @xmath104 , where @xmath105 is the time delay , @xmath106 is the observed radial velocity , @xmath107 is the distance from the continuum source , and @xmath108 is the angle from the line of sight .
the sensitivity of the emissivities ( ew contours in figure 1 ) of the optical recombination lines to the continuum flux , and the resulting consequences to their responsivities , make these lines important candidates for inclusion in the quasar tomography .
horne et al
. found that when applied to the mostly collisionally excited uv emission lines , a certain amount of ambiguity arises in the recovery of the maps due to the weak dependencies in their ews ( and so responsivities ) along lines of constant ionization parameter ( @xmath109 ) corresponding to their maximum emissivities .
their emissivities and responsivities are largely dependent on @xmath110 , with smaller dependencies in gas density due to de - excitation or thermalization .
the optical recombination lines , along with @xmath02800 , offer additional constraints in determining the physical conditions within the blr , in that their emissivities and responsivities are dependent on the local continuum flux and so radius for a fixed central continuum luminosity , as well as changes in this flux as the central continuum source varies .
the response of the optical recombination lines to continuum variations , especially the balmer lines , depends critically on the time - averaged photon flux incident on a spatially extended blr .
this strong dependence exists because of the enhanced excited state populations ( and so optical depths ) that occur at higher continuum flux levels , given the high gas densities found in the blrs of agns .
the emissivities , and hence responsivities , @xmath49 , of these lines are generally strongly anticorrelated with the continuum flux , so @xmath49 changes with distance within the blr for a fixed continuum luminosity and with time as the continuum source varies .
below are the major implications , most of which provide explanations for observed relationships : 1 .
we predict an intrinsic baldwin effect due to this non - linear response , specifically @xmath111 , in most emission lines , including the balmer lines , even after accounting for geometric dilution due to reverberation .
this is a finding that has only recently been observationally confirmed .
regarding the optical recombination lines , this effect should be small for @xmath04686 ( with responsivity @xmath112 ) , but significant for the balmer lines .
we predict that the higher order balmer lines are more responsive , @xmath113 , as is generally observed , and so the broad - line balmer decrement will steepen in low continuum states , and flatten in high states . moreover , @xmath04686 should be substantially more responsive ( @xmath114 ) than the balmer lines , and so large variations in the broad /h@xmath2 integrated flux ratio that are correlated with the continuum state can be expected , even in the absence of a varying continuum sed .
3 . a seyfert 1 galaxy s sub - classification as seyfert 1.0 , 1.5 , 1.8 , or 1.9 may at least in part be due to the relative flux of ionizing photons present within their broad - line regions at the time of observation .
likewise , any given spectrum of a seyfert type 1 agn may pass through some of these sub - classifications as its continuum luminosity fluctuates from high ( seyfert 1 or 1.5 ) to low continuum states ( seyfert 1.8 or 1.9 ) .
the emission - line responsivity , @xmath49 , places an additional important constraint on the run of physical conditions within the blr ( e.g. , gas density , column density , covering fraction ) as a function of radius over and above that of the line emissivity and lag .
this is because for the optical recombination lines , the mean responsivity declines with increasing incident photon flux .
thus , for a particular continuum luminosity state , a smaller emissivity - weighted radius is accompanied by a lower effective responsivity .
greater responsivity in the optical recombination lines is expected during low continuum states , and lower responsivity is expected during high continuum states .
the characteristic size of the blr as measured by the emission - line lag also will track the continuum luminosity level , a phenomenon known as `` breathing . ''
responsivities that are functions of the continuum level and hence time result in non - stationary line transfer functions . in this case
the time dependence is due to changes in the time - averaged photon flux incident on a spatially extended blr
. the stronger response due to an elevated responsivity at low continuum levels will be enhanced by the greater coherence of an effectively smaller blr due to breathing .
this time - dependence in the line transfer functions should also manifest itself in changes in the line profile widths , with the broad - line component becoming narrower / peakier in high continuum states .
because the emissivities and responsivities of the optical recombination lines are anticorrelated with the incident continuum flux , observations that h@xmath3/h@xmath2 and h@xmath2/ flux ratios and the balmer line responsivity decrease from the core to the wings indicate that the blr velocity field diminishes with increasing distance from the central continuum source .
these findings are consistent with recent reverberation studies that find a strong correlation between the emission - line lag and rms profile width for multiple lines in individual agns , which implies that the gas kinematics are dominated by the central massive object .
larger variations in the balmer decrement with continuum state will be found in the broad - line cores rather than in the wings .
observations of such would also indicate that the blr velocity field diminishes with increasing distance from the central continuum source .
differences in broad - line widths ( whether rms or mean profile ) and lags among the balmer and helium recombination emission lines are in large part due to a radial ( continuum flux - dependent ) stratification of optical depth effects in these lines . 9 . much of the observed behavior of the optical recombination lines ( e.g. , the weak response of the balmer line wings ) previously attributed to a substantial contribution from matter - bounded clouds at small blr radii ( i.e. , the so - called vblr ) may be explained by the photon flux dependencies of these lines emissivities and responsivities within ionization - bounded clouds .
these findings should be general , largely independent of any specifically adopted model of the blr . however , the details of the optical recombination lines responsivities will be sensitive to and so diagnostic of specific physical conditions within the blrs of agns .
in addition to the recombination lines investigated here , we suggest that studies of the ir paschen and bracket lines should prove fruitful in further constraining the blr , particularly in its outer regions .
this paper benefited from the careful and critical review of an anonymous referee .
we are grateful to gary ferland for maintaining his freely distributed code , cloudy .
we thank the university of st.andrews and keith horne for their hospitality and support through a pparc visitors grant .
we also thank keith for discussions of work in progress . m.r.g .
would like to thank the hospitality and support of western michigan university during the completion of this work .
appendix : sensitivity of emission line responsivity to model assumptions we briefly discuss the point that most of the major trends in line responsivity reported here should not depend critically on the particular choice of model for the broad emission - line region . as discussed in @xmath4 2 ,
most of these effects can be understood from figure 1 , which shows quite generally the dependence of an emission line s emissivity on cloud density and incident ionizing photon flux , irrespective of how these clouds ( or emitting entities ) are distributed or other details .
korista et al . ( 1997 ) discusses and illustrates the effects of cloud column density on the line emissivities , and thus ew contours in figure 1 .
the present simulations assume a constant cloud column density of @xmath12 @xmath13 . in general ,
the computed integrated agn line spectrum is relatively insensitive to this parameter for column densities exceeding @xmath115 @xmath13 ( see also goad & koratkar 1998 ) , except for the high ionization lines of @xmath01034 and @xmath0774 , which require column densities @xmath116 @xmath13 and @xmath117 @xmath13 to fully form within the cloud ( korista et al .
1997 ; hamann et al .
1998 ) . to the extent that cloud column density has a significant effect on the luminosities or responsivities of the emission lines plotted in figure 1
, the clouds must lie in the vicinity of the central diagonal boundary representing the steep downward gradient in ew ( i.e. , near @xmath118 )
. these would be the clouds affected by the passage of an ionization front .
if the preponderance of clouds have a smaller column density than those computed here , this _ diagonal _ boundary would shift toward the lower right corners of the density - flux planes in figure 1 . for higher column density clouds
, this boundary would shift toward the upper left .
the emissivities of clouds lying far from this boundary in the density - flux plane would be essentially unaffected .
this is clearly illustrated in figure 3 of korista et al .
( 1997 ) . from figure 4 and eq . 2 in the present paper
, it can therefore be seen that changing the cloud column density by an order of magnitude in either direction from the adopted value will not change the overall trends of the ew and responsivity of the ionization - bounded clouds being proportional to the incident continuum flux for the four emission lines studied here .
because of their inefficient natures , matter - bounded clouds can not contribute substantially to the hydrogen recombination line luminosity ( or responsivity ) , unless they have much larger covering fractions than the optically thick clouds at the same radius .
even then , their importance is likely to be minor , except perhaps in the extreme wings given a velocity field that declines with distance from the central source .
different assumptions about the incident continuum sed or gas abundances do not change the line ew distributions in figure 1 very much ( see korista et al .
1997 ) , and so would have little effect on these lines responsivities .
if the optical depths in the balmer lines were significantly reduced by the presence of large extra - thermal velocities that might occur in an accretion disk wind ( e.g. , murray & chiang 1997 ) , the result would be the ew contours in figure 1 peaking up at higher incident fluxes and densities and the balmer line - emitting region being more strongly influenced by gas with high responsivity ( @xmath119 ) .
since the balmer lines have reported responsivities less than 1 even after correcting for geometric dilution , this would constrain models with large extra - thermal velocities and/or balmer emission from large outer radii . other consequences of significant extra - thermal velocities might be smaller predicted lags and broader predicted line profiles ( for a fixed central mass and incident continuum luminosity ) . if in general the radiative transfer of the balmer lines is currently less than adequately handled ( as may be the case ; see netzer et al .
1995 ) , the optical depth of h@xmath3 should nevertheless remain greater than that of h@xmath2 , and those of the helium lines should remain smaller than the balmer lines .
thus , the relative senses of the ew contours of the balmer lines in figure 1 should be preserved .
the sense of the balmer decrement variability as a function of continuum state should also be preserved , as should the h@xmath3/h@xmath2 intensity ratio across the line profile : _ provided that the bulk of the velocity field diminishes with increasing distance from the central source_.
most of the extra responsivity in h@xmath2 over h@xmath3 in figure 7 lies within the line core whose contributions are dominated by gas lying ( and responding ) at larger distances . as a consequence , the difference in the measured lags of these two lines may not be as large as their ew distributions in figure 1 might indicate .
that is , their responsivity - weighted radii may be closer in size than their emissivity - weighted radii ( goad et al .
in fact , depending on the actual emissivity and responsivity distributions of the two lines , it is possible for h@xmath2 to lag slightly behind h@xmath3 .
in other words the total light profile may be broader in h@xmath2 than h@xmath3 and the other way around in the rms variation profile , with h@xmath3 having a slightly smaller lag .
the small differences in the balmer lines beaming functions ( obrien et al .
1994 ) may also play a role in altering the relative responsivity - weighted radii , and thus relative lags of these emission lines .
gondhalekar , p.m. 1987 , in `` emission lines in active galactic nuclei , '' proceedings of a workshop on astronomy and astrophysics ( chilton , england : rutherford appleton laboratory ) , ed .
p.m. gondhalekar lccccc & & & & & + h@xmath3 @xmath06563 & 0.52 & 0.43 & 0.63 & 4200 & 4600 + h@xmath2 @xmath04861 & 0.64 & 0.54 & 0.77 & 4510 & 5230 + h@xmath11 @xmath04340 & 0.68 & 0.58 & 0.79 & 4570 & 5250 + @xmath05876 & 0.83 & 0.77 & 0.87 & 5780 & 7350 + @xmath04686 & 0.92 & 0.87 & 0.97 & 7590 & 9480 + ly@xmath3 @xmath01216 & 0.74 & 0.71 & 0.77 & 5180 & 5280 + | we investigate the effect of a global change in the ionizing continuum level on the behavior of the strong optical broad emission lines seen in spectra of the nuclear emission - line regions of active galactic nuclei ( agn ) , including the balmer lines , @xmath05876 , and @xmath04686 . unlike most of the prominent heavy element lines found in the uv , the optical hydrogen and helium recombination lines emissivities are strongly dependent on the incident continuum flux , since these lines arise out of excited states whose _ optical depths depend on the incident flux of photons_. using photoionization calculations we determine the luminosity - dependent responsivities , @xmath1 , of these lines for a general model of the broad emission line region ( blr ) , with the purpose of establishing them as important probes of the physical conditions within the blr of agns .
the dependence of these lines emissivities on the incident photon flux invokes a dependence in their responsivities on distance from the central continuum source .
in particular , the responsivities of these lines are generally anticorrelated with the incident photon flux . thus
, their responsivities vary with distance within the blr for a fixed continuum luminosity and change with time as the continuum source varies . consequently , after correcting for light - travel - time effects the response of the balmer and optical helium lines should generally be strongest during low continuum luminosity states .
responsivity that depends on photon flux and continuum state may explain a number of outstanding problems currently under investigation in broad - line variability studies of these and other emission lines .
these include the origin of the intrinsic baldwin effect , measurements of luminosity - dependent lags ( a `` breathing '' blr ) and luminosity - dependent variations in : integrated broad emission - line flux ratios ( including @xmath04686/h@xmath2 ) , broad line profile shapes , and radial velocity - dependent intensity ratios .
the broad h@xmath3/h@xmath2 and h@xmath2/ flux ratios and the balmer emission - line responsivity are observed to decrease from the line center to the line wings .
these , along with our findings , lead to the conclusion that the blr velocity field diminishes with increasing distance from the central continuum source .
this is consistent with recent reverberation studies that find a relationship between the emission - line lag and rms profile width for multiple lines in individual agn , which implies that the velocity field is dominated by a central massive object .
finally , the responsivity of ionization - bounded clouds can account for much of the observed behavior of the optical recombination lines ( e.g. , the weak response of the balmer line wings ) previously attributed to a substantial contribution from matter - bounded clouds at small blr radii . |
recent observations of cosmic microwave background ( cmb ) anisotropies as well as density perturbations in the large - scale structure strongly support the basic predictions of inflationary scenarios , in which primordial adiabatic fluctuations were produced during the accelerated phase of the cosmic expansion and their statistical properties are approximately described by gaussian statistics with a nearly scale - invariant power spectrum ( e.g. , @xcite ) .
some specific inflationary models have been ruled out by observations , narrowing the constraints on the early stage of the universe . with precision measurements from future observations
, we will detect clear signals that help us to discriminate between many candidate of inflationary models . amongst several signals accessible in the near future , departure from gaussianity
is an important clue to probe the generation mechanism for primordial perturbations as well as to discriminate between inflation models .
while the simplest slow - roll inflation with single scalar field predicts a small departure from gaussianity ( e.g. , @xcite ) , models with a non - trivial kinetic term or late - time inflaton decay called curvaton scenario @xcite can produce a large non - gaussianity ( e.g. , @xcite ) .
there are also viable models motivated by string theory that produce large non - gaussianity such as dirac - born - infeld ( dbi ) inflation @xcite and ekyprotic scenario @xcite .
although tentative detections of primordial non - gaussianity have been reported very recently ( @xcite , but see also @xcite ) and the result is broadly consistent with the standard prediction of slow - roll inflation , these detections are at relatively low statistical significance and more precise measurements are necessary for a definite detection of the non - gaussian signals .
planned cmb surveys , namely the planck mission @xcite , will have much greater sensitivity and better detect non - gaussian signals ( e.g. , @xcite ) .
further , large - scale structure data from future spectroscopic surveys will uncover the mass density fluctuations up to giga parsec scales , which preserve the statistical properties of primordial fluctuations ( e.g. , @xcite ) .
since these two measurements can probe different scales of primordial fluctuations , they are , in principle , complementary to each other . in the present paper ,
we study the signature of primordial non - gaussianity imprinted on the large - scale structure , especially focusing on the matter power spectrum .
naively , we expect that the signature of primordial non - gaussianity basically appears in the statistical properties of higher - order quantities such as the bispectrum and trispectrum , and the power spectrum as a second - order statistic remains unchanged even if large non - gaussianity has been produced .
however , this is true only when the fluctuations of the mass density field are tiny and well within the linear regime .
as the gravitational clustering develops , the coupling between different fourier modes becomes important and the scale - dependent non - linear growth appears due to the mode - coupling effect . in the weakly non - linear regime ,
the strength of this mode - coupling sensitively depends on the initial condition .
since the non - gaussian density field intrinsically possesses non - trivial mode - correlations , the evolution of the power spectrum may be altered by the primordial non - gaussianity , at least in the weakly non - linear regime . here ,
employing the perturbative calculations , we study the effect of primordial non - gaussianity on the matter power spectrum . the effect of non - gaussianity on the matter power spectrum has been previously investigated by ref.@xcite using perturbation theory , showing that the magnitude of this effect is roughly at the few percent level in the weakly non - linear regime . in this paper , adopting the two representative models of primordial non - gaussianity described in sec .
[ sec : nongaussian_model ] , we quantitatively estimate the non - gaussian effects on the power spectrum and elucidate their shape dependence ( sec . [ sec : non_gaussianpt ] ) .
in particular , we carefully examine the influence of primordial non - gaussianity on cosmological parameter estimation , paying particular attention to the primordial spectral indices and the characteristic scale of the baryon acoustic oscillations as a cosmic standard ruler ( sec .
[ sec : fisher ] ) . also , we discuss the non - gaussian effect on the galaxy biasing ( sec .
[ sec : galaxy_biasing ] ) .
this issue has been recently studied by several authors in the context of the halo biasing prescription @xcite , and local biasing prescription @xcite . here
, we adopt the local biasing scheme and consider the scale - dependent galaxy biasing arising from the primordial non - gaussianity . throughout the paper ,
we assume a flat lambda cold dark matter ( cdm ) model and the fiducial model parameters are chosen based on the five - year wmap results ( wmap+bao+sn ml , see ref.@xcite ) : @xmath0 , @xmath1 , @xmath2 , @xmath3 , @xmath4 , @xmath5 , @xmath6 .
according to the commonly accepted mechanisms to generate the primordial fluctuations , non - gaussianity would be imprinted on the primordial curvature perturbation produced during inflation .
the curvature perturbation can also be generated after inflation by a late - time decay of light fields , and this would produce a large non - gaussianity in the curvature perturbation .
further , as alternative scenarios to inflation such as ekpyrotic models , nearly scale - invariant curvature perturbation with large non - gaussianity could be generated during the collapsing phase of the universe .
let us denote the primordial curvature perturbation on the uniform density hypersurface and the bardeen s curvature potential at super - horizon scales by @xmath7 and @xmath8 , respectively .
the primordial density perturbation defined at an early time of the matter dominated epoch , @xmath9 , is related to these variables through ( e.g. , @xcite ) @xmath10 with the function @xmath11 being @xmath12 where the function @xmath13 is the transfer function of matter fluctuations normalized to unity at @xmath14 .
note that the curvature potential given in the matter dominated epoch , @xmath8 , is related to the curvature perturbation @xmath15 . using the relation ( [ eq : delta0_phi_p ] )
, @xmath16-point statistics of the primordial density field are all expressed in terms of respective correlator of the curvature potential or the curvature perturbation .
for the power spectra , we have @xmath17 with quantities @xmath18 and @xmath19 defined by @xmath20 the bispectrum as the first non - trivial quantity arising from the non - gaussianity becomes @xmath21 where we define @xmath22 details of the functional form of @xmath23 depend on the generation mechanism of primordial non - gaussianity . in the following ,
we will consider two representative models of non - gaussianity and give the explicit expressions for the primordial bispectrum @xmath23 . in the local model of primordial non - gaussianity
, the primordial fluctuation @xmath8 is characterized by the taylor expansion of the gaussian field .
denoting the gaussian field by @xmath24 , it is conventionally characterized as : @xmath25 the fourier counterpart of the above equation is given by @xmath26 alternatively , we can expand @xmath7 as : @xmath27 where @xmath28 is gaussian field and we have the relation @xmath29 . for the cases of interest here , deviation from gaussianity
is generally small and perturbative evaluation of the primordial bispectrum is valid .
the leading - order result becomes @xmath30 .
\label{eq : bispectrum_local}\end{aligned}\ ] ] in the slow - roll inflation models , @xmath31 is suppressed by the slow - roll parameters and thus it is expected to be very small . however , in curvaton models or ekpyrotic models , it is possible to generate a large non - gaussianity of the local model with @xmath32 . as another type of bispectrum configuration
, we consider an equilateral model .
the dbi inflation is a typical example where the bispectrum has a maximum amplitude for equilateral configurations .
this is in contrast with the local model where the bispectrum has the largest amplitude for squeezed configurations @xcite .
the bispectrum in the equilateral model is approximated by @xcite . ]
@xmath33 , \label{eq : bispectrum_eq}\end{aligned}\ ] ] with @xmath34 . here
, we introduce the slope index @xmath35 in order to characterize the scale - dependence of the non - gaussianity , which generically appears in the dbi - type inflation models .
we set the characteristic scale to @xmath36 .
note that the bispectrum in equation ( [ eq : bispectrum_eq ] ) is normalized in such a way that , for equilateral configurations ( @xmath37 ) , it coincides with the local form given in equation ( [ eq : bispectrum_local ] ) .
even if the primordial fluctuations are well described by linear theory , the non - linearity of the gravitational dynamics eventually dominates and we must correctly take into account the non - linear growth of matter fluctuations , which significantly modifies the matter power spectrum . for the scales of interest here , especially the accessible scales of future galaxy redshift surveys , the non - linear evolution is rather moderate and perturbative treatment is valid .
the mass density field is described by @xmath38 in the above , the function @xmath39 represents the linear fluctuation and it is given by @xmath40 , where @xmath41 is the linear growth rate .
neglecting the decaying mode of linear perturbation , the solutions for higher - order quantities are formally expressed as @xcite @xmath42^n \,\int\frac{d^3k_1\cdots d^3k_n } { ( 2\pi)^{3(n-1)}}\,\delta_{\rm d}({\mbox{\boldmath$k$}}-{\mbox{\boldmath$k$}}_1-\cdots-{\mbox{\boldmath$k$}}_n)\ , f^{(n)}_{\rm sym}({\mbox{\boldmath$k$}}_1,\cdots,{\mbox{\boldmath$k$}}_n)\delta_0({\mbox{\boldmath$k$}}_1)\cdots\delta_0({\mbox{\boldmath$k$}}_n ) , \label{eq : pt_sol}\end{aligned}\ ] ] with @xmath43 being the symmetrized kernel for the @xmath16-th order solution .
then , the power spectrum of the density field , defined as @xmath44 can be calculated by substituting eq.([eq : pt_sol ] ) into the above expression .
the resultant expression is summarized as @xmath45\,+\cdots,\ ] ] where the terms @xmath46 , @xmath47 and @xmath48 are the so - called one - loop power spectra given by @xmath49 these are the most general expressions for the one - loop power spectra in the presence of primordial non - gaussianity .
the quantity @xmath46 is the first non - trivial correction associated with the primordial density bispectrum @xmath50 .
while the quantities @xmath47 and @xmath48 represent the first leading - order corrections in the case of gaussian initial conditions , additional contributions originating from the primordial density trispectrum , @xmath51 , also arise .
thus , in general , we need the explicit functional form of the primordial trispectrum as well as primordial bispectrum in order to compute the one - loop power spectra .
however , in most general cases , the primordial trispectrum is of the order of @xmath52 , which is negligibly smaller than the bispectrum , @xmath53 .
since the terms including @xmath54 may be regarded as the two - loop order , we can safely neglect the trispectrum contribution . with this treatment ,
the terms @xmath47 and @xmath48 reduce to nothing but the _ standard _ one - loop spectra , and they are explicitly given by ( e.g. , @xcite ) @xmath55 ^ 2 , \label{eq : p_ab^22(k ) } \\
p^{(13)}(k;z)&=&d^4(z)\,\frac{k^3}{(2\pi)^2}\,\,p_0(k)\ , \int_0^{\infty } dx\ , x^2 p_0(kx ) \nonumber\\ & & \quad\quad\times\,\,\frac{1}{252\,x^2 } \left[\frac{12}{x^2}-158 + 100x^2 - 42x^4 + \frac{3}{x^3}\,(x^2 - 1)^3(7x^2 + 2)\,\ln\left|\frac{x+1}{x-1}\right| \right ] .
\label{eq : p_ab^13(k)}\end{aligned}\ ] ] also , the non - trivial contribution from the primordial bispectrum is expressed in terms of the bispectrum of the curvature perturbation as @xmath56 ( right ) , for @xmath57 .
the smooth spectra are obtained from the no - wiggle approximation of the linear transfer function according to ref .
the lines from top to bottom respectively indicate @xmath58 ( solid ) , @xmath59 ( long - dashed ) , @xmath60 ( dotted ) , @xmath61 ( short - dashed ) and @xmath62 ( dot - dashed ) . [ fig : ratio_nowiggle],title="fig:",width=321 ] ( right ) , for @xmath57 .
the smooth spectra are obtained from the no - wiggle approximation of the linear transfer function according to ref .
the lines from top to bottom respectively indicate @xmath58 ( solid ) , @xmath59 ( long - dashed ) , @xmath60 ( dotted ) , @xmath61 ( short - dashed ) and @xmath62 ( dot - dashed ) .
[ fig : ratio_nowiggle],title="fig:",width=321 ] , for @xmath63 ( left ) , @xmath64 ( middle ) and @xmath65 ( right ) . from top to bottom panels , the results for the local , equilateral with @xmath66 , and @xmath67 are shown , respectively . in each panel , we plot the cases with non - gaussian parameter @xmath68 ( solid ) , @xmath69 ( long - dashed ) , @xmath70 ( short - dashed ) and @xmath71 ( dot - dashed ) .
the vertical arrows labeled by @xmath72 and @xmath73 indicate the maximum wave number below which the perturbation theory predictions are reliable with a precision of @xmath74 and @xmath75 level , respectively , according to the criteria ( [ eq : criteria ] ) @xcite . as references ,
the error bars limited by the cosmic variance are plotted in the right panel , assuming the survey volume of @xmath76gpc@xmath77 ( thin solid ) and @xmath78gpc@xmath77 ( thick solid ) .
[ fig : ratio_pk],width=396 ] , for @xmath63 ( left ) , @xmath64 ( middle ) and @xmath65 ( right ) . from top to bottom panels , the results for the local , equilateral with @xmath66 , and @xmath67 are shown , respectively . in each panel , we plot the cases with non - gaussian parameter @xmath68 ( solid ) , @xmath69 ( long - dashed ) , @xmath70 ( short - dashed ) and @xmath71 ( dot - dashed ) .
the vertical arrows labeled by @xmath72 and @xmath73 indicate the maximum wave number below which the perturbation theory predictions are reliable with a precision of @xmath74 and @xmath75 level , respectively , according to the criteria ( [ eq : criteria ] ) @xcite . as references ,
the error bars limited by the cosmic variance are plotted in the right panel , assuming the survey volume of @xmath76gpc@xmath77 ( thin solid ) and @xmath78gpc@xmath77 ( thick solid ) .
[ fig : ratio_pk],width=396 ] , for @xmath63 ( left ) , @xmath64 ( middle ) and @xmath65 ( right ) . from top to bottom panels , the results for the local , equilateral with @xmath66 , and @xmath67 are shown , respectively . in each panel , we plot the cases with non - gaussian parameter @xmath68 ( solid ) , @xmath69 ( long - dashed ) , @xmath70 ( short - dashed ) and @xmath71 ( dot - dashed ) .
the vertical arrows labeled by @xmath72 and @xmath73 indicate the maximum wave number below which the perturbation theory predictions are reliable with a precision of @xmath74 and @xmath75 level , respectively , according to the criteria ( [ eq : criteria ] ) @xcite . as references ,
the error bars limited by the cosmic variance are plotted in the right panel , assuming the survey volume of @xmath76gpc@xmath77 ( thin solid ) and @xmath78gpc@xmath77 ( thick solid ) .
[ fig : ratio_pk],width=396 ] , for @xmath63 ( left ) , @xmath64 ( middle ) and @xmath65 ( right ) . from top to bottom panels , the results for the local , equilateral with @xmath66 , and @xmath67 are shown , respectively . in each panel
, we plot the cases with non - gaussian parameter @xmath68 ( solid ) , @xmath69 ( long - dashed ) , @xmath70 ( short - dashed ) and @xmath71 ( dot - dashed ) . the vertical arrows labeled by @xmath72 and @xmath73 indicate the maximum wave number below which the perturbation theory predictions are reliable with a precision of @xmath74 and @xmath75 level , respectively , according to the criteria ( [ eq : criteria ] ) @xcite . as references ,
the error bars limited by the cosmic variance are plotted in the right panel , assuming the survey volume of @xmath76gpc@xmath77 ( thin solid ) and @xmath78gpc@xmath77 ( thick solid ) . [
fig : ratio_pk],width=396 ] based on the expressions given above , we now calculate the matter power spectrum , focusing on the scales relevant for future galaxy surveys .
figure [ fig : ratio_nowiggle ] plots the ratios of power spectra to the smooth reference spectrum , @xmath79 , given at @xmath57 .
the reference spectrum @xmath80 is the linear power spectra computed from the no - wiggle approximation of the transfer function given by ref .
the left panel shows the results for the local model , and the right panel plots the results for the equilateral model with index @xmath81 .
the overall behavior at higher @xmath82 is basically dominated by the corrections coming from the standard 1-loop power spectra , @xmath47 and @xmath48 , but small variation of the power spectrum amplitudes manifests at higher @xmath82 .
this is the contribution of primordial non - gaussianity , @xmath46 , originating from the primordial bispectrum .
depending on the sign of @xmath31 , the primordial non - gaussianity enhances or suppresses the non - linear growth of the power spectrum and the deviation from the gaussian result @xmath83 becomes more significant as @xmath84 increases .
although the effect of primordial non - gaussianity seems rather mild within the current constrained values of @xmath85 and @xmath86 ( 95%c.l.)@xcite , we do not immediately exclude the possibility of significant impact on the cosmological parameter estimation .
we will discuss this issue in the next section .
next we will focus on the shape of the power spectrum . in figure
[ fig : ratio_pk ] , the ratios of the power spectra to those in the gaussian case ( @xmath60 ) , i.e. , @xmath87 , are plotted for various redshifts and values of @xmath31 .
note that in computing the ratio , both the numerator and denominator are calculated from perturbation theory . in figures [ fig : ratio_nowiggle ] and [ fig :
ratio_pk ] , the vertical arrows labeled by @xmath72 and @xmath73 indicate the maximum wavenumber below which the precision level of perturbation theory prediction is expected to be better than @xmath74 and @xmath75 , respectively . according to recent numerical experiments , the maximum wavenumber is empirically determined through @xcite @xmath88 where @xmath89 for @xmath90 . solving the above equation with respect to @xmath82
, we obtain @xmath72 and @xmath73 as function of redshift .
note that the maximum wave numbers given above have been derived by comparison between n - body simulations and theoretical predictions , and the resultant convergence range is narrower than those previously proposed ( e.g. , @xcite ) . in this sense , eq.([eq : criteria ] ) may be regarded as a conservative criterion .
bearing the limitation of perturbation theory in mind , we see that the deviation of the power spectrum from the gaussian case also depends on the redshift and scales as well as the model of primordial non - gaussianity .
monotonic change of the power spectrum amplitude is broadly consistent with recent n - body simulations in the case of the local model @xcite , although the fractional change is typically @xmath91 , as expected from figure [ fig : ratio_nowiggle ] .
note , however , that the level of this changes roughly corresponds to the sensitivity achievable with future redshift surveys . as references ,
we plot the error bars in each right panel , which represent the expected @xmath92 errors , @xmath93 , limited by the cosmic - variance , i.e. , @xmath94 with @xmath95 being the number of fourier modes within a given bin at @xmath82 .
here , we specifically consider the two representative cases of @xmath96gpc@xmath77 ( thin ) and @xmath97gpc@xmath77 ( thick ) , corresponding to the future surveys dedicated for the measurement of baryon acoustic oscillations ( baos ) from the ground and space , respectively
. naively comparing the error bars to the amplitudes of the non - gaussian effects , ground - based bao surveys with a typical volume of @xmath98few @xmath99gpc@xmath100 will find it difficult to detect the signature of primordial non - gaussianity only from the power spectrum .
however , idealistic surveys with huge volumes seem to show the potential for a definite detection of non - gaussian effects even with the currently constrained values of @xmath31 .
apart from primordial non - gaussianity , there are several parameters that affect the shape of the power spectrum . among these ,
the primordial spectral index @xmath101 and running index @xmath102 monotonically change the power spectrum , which resemble the effect of primordial non - gaussianity at some wavenumbers .
a natural question arises whether the primordial non - gaussianity is indeed detectable or not considering the degeneracies and how the non - negligible effect of non - gaussianity adversely affects the estimations of primordial spectral index and the running of the index . on the other hand , for the scales accessible to the future surveys ,
a measurement of the characteristic scale of baos is an important clue to probe the late - time acceleration of the universe e.g. , @xcite ) , and a percent - level determination of the acoustic scale is required for the determination of the dark energy equation of state @xcite . in this respect ,
even the small effect of primordial non - gaussianity may affect the determination of the bao scale . here
, we apply the fisher - matrix method to address these issues and explore the potential impact on cosmological parameter estimation as well as the determination of the bao scale . for this purpose
, we consider a rather simplified setup , namely that our observable is the real - space power spectrum free from the redshift - space distortion . under the assumption of linear galaxy biasing
, the observed power spectrum may be written as ( e.g. , @xcite ) : @xmath103 where the power spectrum @xmath104 is computed from perturbation theory . here , @xmath105 is the linear biasing parameter , and @xmath106 represents the cosmological distance averaged over three - dimensional space , @xmath107^{1/3}$ ] .
the subscript `` true '' denotes the quantities estimated from the true cosmological model .
following refs.@xcite , the fisher matrix for the galaxy power spectrum is given by @xmath108 where @xmath109 represents one from a set of parameters .
the quantity @xmath110 is the survey volume and @xmath111 is the number density of galaxies .
the range of integration @xmath112 $ ] should be determined through the survey properties , and in particular the minimum wave number is limited to @xmath113 .
note that the choice of @xmath114 may be crucial for the power spectrum measurement on large - scales @xmath115mpc@xmath116 , where the scale - dependent effect of galaxy biasing becomes prominent through the non - trivial mode - coupling from primordial non - gaussianity @xcite . here , we fix @xmath117mpc@xmath116 and allow @xmath118 to vary .
the effect of scale - dependent biasing will be discussed in next section . in the fisher - matrix analysis presented below
, we include the five parameters given by @xmath119 .
the fiducial values of these parameters are set to @xmath3 , @xmath120 , @xmath121 , and @xmath60 .
we then consider a space - based bao survey as an idealistically gigantic galaxy survey along the lines of space and adept ( @xcite , see also ref .
we adopt the survey parameters of a space - based bao experiment as follows : @xmath122 , @xmath123(@xmath124gpc)@xmath100 , @xmath125 , and @xmath126(@xmath124mpc)@xmath127 ( e.g. , @xcite ) .
below , we mainly focus on the local model of primordial non - gaussianity .
the results for the equilateral model are qualitatively the same as in the local model and are briefly summarized in table [ tab : fisher_results ] . and @xmath101 ( left ) , @xmath128 ( middle ) , and @xmath129 ( right ) , assuming the local model of primordial non - gaussianity .
in deriving the constraints , we fix the maximum wavenumber to @xmath130mpc@xmath116 and adopt the survey parameters of a space - based bao mission .
the different lines indicate the results from imposing different prior information on @xmath101 and @xmath128 : no priors ( solid ) : @xmath131 ( short - dashed ) : @xmath131 , @xmath132 ( long - dashed ) .
[ fig:2derror_ns_nrun_dv_adept],title="fig:",width=215 ] and @xmath101 ( left ) , @xmath128 ( middle ) , and @xmath129 ( right ) , assuming the local model of primordial non - gaussianity . in deriving the constraints ,
we fix the maximum wavenumber to @xmath130mpc@xmath116 and adopt the survey parameters of a space - based bao mission .
the different lines indicate the results from imposing different prior information on @xmath101 and @xmath128 : no priors ( solid ) : @xmath131 ( short - dashed ) : @xmath131 , @xmath132 ( long - dashed ) .
[ fig:2derror_ns_nrun_dv_adept],title="fig:",width=215 ] and @xmath101 ( left ) , @xmath128 ( middle ) , and @xmath129 ( right ) , assuming the local model of primordial non - gaussianity . in deriving the constraints
, we fix the maximum wavenumber to @xmath130mpc@xmath116 and adopt the survey parameters of a space - based bao mission .
the different lines indicate the results from imposing different prior information on @xmath101 and @xmath128 : no priors ( solid ) : @xmath131 ( short - dashed ) : @xmath131 , @xmath132 ( long - dashed ) .
[ fig:2derror_ns_nrun_dv_adept],title="fig:",width=215 ] and @xmath101 ( left ) , @xmath128 ( middle ) , and @xmath129 ( right ) , assuming the local model of primordial non - gaussianity . in deriving the constraints , we fix the maximum wavenumber to @xmath130mpc@xmath116 and adopt the survey parameters of a space - based bao mission .
the different lines indicate the results from imposing different prior information on @xmath101 and @xmath128 : no priors ( solid ) : @xmath131 ( short - dashed ) : @xmath131 , @xmath132 ( long - dashed ) .
[ fig:2derror_ns_nrun_dv_adept],title="fig:",height=192 ] and @xmath101 ( left ) , @xmath128 ( middle ) , and @xmath129 ( right ) , assuming the local model of primordial non - gaussianity . in deriving the constraints
, we fix the maximum wavenumber to @xmath130mpc@xmath116 and adopt the survey parameters of a space - based bao mission .
the different lines indicate the results from imposing different prior information on @xmath101 and @xmath128 : no priors ( solid ) : @xmath131 ( short - dashed ) : @xmath131 , @xmath132 ( long - dashed ) .
[ fig:2derror_ns_nrun_dv_adept],title="fig:",height=192 ] and @xmath101 ( left ) , @xmath128 ( middle ) , and @xmath129 ( right ) , assuming the local model of primordial non - gaussianity . in deriving the constraints
, we fix the maximum wavenumber to @xmath130mpc@xmath116 and adopt the survey parameters of a space - based bao mission .
the different lines indicate the results from imposing different prior information on @xmath101 and @xmath128 : no priors ( solid ) : @xmath131 ( short - dashed ) : @xmath131 , @xmath132 ( long - dashed ) .
[ fig:2derror_ns_nrun_dv_adept],title="fig:",height=192 ] .[tab : fisher_results]marginalized one - dimensional errors ( 68%c.l . ) on @xmath101 , @xmath128 , @xmath129 and non - gaussian parameters @xmath31 and @xmath35 from fisher - matrix analysis adopting the survey parameters of a space - based bao experiment [ cols="<,^,^,^,^,^",options="header " , ] figure [ fig:1derror_ns_nrun_dv_adept ] shows the marginalized 1-@xmath133 ( 68%c.l . ) errors on the spectral index @xmath101 ( left ) , the running of the index @xmath128 ( middle ) , and the distance scale @xmath129 ( right ) as function of cutoff scale , @xmath118 . here
, the solid lines represent the results assuming no prior information to these parameters ( _ prior 1 _ , solid ) .
the short- and long - dashed lines correspond to the results taking account of the priors on @xmath101 and @xmath128 , which are expected to come from the upcoming planck cmb experiment : @xmath131 ( _ prior 2 _ , short - dashed ) ; @xmath131 and @xmath134 ( _ prior 3 _ , long - dashed ) .
we assume gaussian priors for these cases .
figure [ fig:1derror_ns_nrun_dv_adept ] indicates that the cmb priors play an important role for the determination of the primordial indices and distance scales .
this has been repeatedly pointed out in the literature , but we emphasize that it is indeed crucial when we include the non - gaussian parameter @xmath31 in the analysis of cosmological parameter estimation . to clarify the influences , we exclude @xmath31 from the fisher - matrix analysis and evaluate the marginalized errors again .
the results are plotted in thin lines . comparing thick with thin lines
, we see that the inclusion of the extra parameter @xmath31 significantly worsen the constraints on the primordial index and its running . with strong priors on @xmath101 and @xmath128 ( _ prior 3 _ )
, tighter constraints comparable to those ignoring primordial non - gaussianity would be obtained . by contrast , the determination of distance scale @xmath129 is insensitive to primordial non - gaussianity if we choose the cutoff wavenumber @xmath118 around @xmath72 , labeled by the vertical arrows .
this is partly because the distance scale is mainly determined through the baos at linear scales , where the effect of primordial non - gaussianity becomes negligible at @xmath135 .
it is interesting to note that all the curves depicted in figure [ fig:1derror_ns_nrun_dv_adept ] become almost constant at @xmath136mpc@xmath116 .
since this scale is slightly larger than the one inferred from the mean separation of galaxies ( @xmath137mpc@xmath116 ) , the constancy of the predicted errors implies that the shot - noise contribution becomes significant at @xmath138 . in figure [ fig:2derror_ns_nrun_dv_adept ] , we next plot the two - dimensional constraints on the non - gaussian parameter @xmath31 and one of the free parameters @xmath101 , @xmath128 and @xmath129 , marginalizing over the other remaining parameters . in this plot ,
the cutoff scale is fixed to @xmath139 , and the 1-@xmath133 ( 68%c.l . )
contours are shown for the three different priors . as expected from figure [ fig:1derror_ns_nrun_dv_adept ] , the primordial index @xmath101 and its running @xmath128 without cmb priors ( _ prior 1 _ ) exhibit a strong degeneracy with the @xmath31 parameter , while the degeneracy between @xmath31 and distance scale is very weak .
the uncertainty in the @xmath31 parameter is therefore very large in this case .
the errors can be reduced significantly when the strong constraints on @xmath101 and @xmath128 are obtained ( _ prior 3 _ ) . however , the uncertainty in @xmath31 is still @xmath140 , which is quite a bit larger than other observational techniques using the bispectrum or cluster abundance @xcite .
the uncertainty may be reduced if we increase the cutoff scale @xmath118 , although it is generally difficult to accurately predict the power spectrum at these scales , including the higher - order corrections . hence
, it is difficult to detect the non - gaussian signals even from an idealistic large - volume survey .
table [ tab : fisher_results ] summarizes the results of the one - dimensional marginalized errors for various cases with different fiducial values , including the case of the equilateral model . here
, in addition to the priors on @xmath101 and @xmath128 ( _ prior 3 _ ) , we further put a gaussian prior on @xmath35 with @xmath141 , the reason for which comes from the fact that @xmath35 is the slow - roll parameter and should be restricted to @xmath142 . because of the additional parameter , the resultant errors on @xmath31 significantly increase for the equilateral model , especially with @xmath143 .
by contrast , the errors on @xmath101 , @xmath128 and distance scale are hardly affected .
finally , as a representative example for a more near - term project , we consider the case of a ground - based bao survey .
the survey parameters we adopt are @xmath57 , @xmath144(@xmath124gpc)@xmath100 , @xmath145 , and @xmath146 ( @xmath124mpc)@xmath127 ( @xcite , see also @xcite ) .
figure [ fig : errors_dv_wfmos ] shows the expected one - dimensional error on @xmath129 ( left ) and the two - dimensional errors on @xmath31 and @xmath129 ( right ) .
compared to the large volume of a space - based bao experiment , the statistical errors of the measurement of the power spectrum are somewhat larger and thereby the precision of the distance - scale estimation is rather degraded .
nevertheless , the results are hardly affected by the primordial non - gaussianity . in conclusion ,
the presence of primordial non - gaussianity potentially affects the estimation of the primordial index and the running , and the cmb - based priors on these parameters would be crucial , but the distance - scale measurement is relatively insensitive to it for the appropriate choice of @xmath118 . in the end , the primordial non - gaussianity is hardly detectable from galaxy redshift surveys even with idealistically large survey - volume .
( 68%c.l . ) error on the distance scale as a function of maximum wavenumber ( left ) and two - dimensional joint @xmath147 c.l .
constraints on @xmath31 of the local model and @xmath129 fixing the maximum wavenumber to @xmath148mpc@xmath116 ( right ) . here
, we specifically adopt the survey parameters of @xmath57 , @xmath144@xmath99gpc@xmath100 , @xmath149 , and @xmath146 @xmath150mpc@xmath127 , as a representative example of ground - based bao surveys .
the meaning of the lines are the same as in figures [ fig:1derror_ns_nrun_dv_adept ] and [ fig:2derror_ns_nrun_dv_adept ] .
[ fig : errors_dv_wfmos],title="fig:",height=245 ] ( 68%c.l . )
error on the distance scale as a function of maximum wavenumber ( left ) and two - dimensional joint @xmath147 c.l .
constraints on @xmath31 of the local model and @xmath129 fixing the maximum wavenumber to @xmath148mpc@xmath116 ( right ) . here
, we specifically adopt the survey parameters of @xmath57 , @xmath144@xmath99gpc@xmath100 , @xmath149 , and @xmath146 @xmath150mpc@xmath127 , as a representative example of ground - based bao surveys .
the meaning of the lines are the same as in figures [ fig:1derror_ns_nrun_dv_adept ] and [ fig:2derror_ns_nrun_dv_adept ] .
[ fig : errors_dv_wfmos],title="fig:",height=245 ]
so far , we have mainly dealt with the matter power spectrum and the galaxy power spectrum has been only considered under the simplified assumption of linear galaxy biasing . in reality , the statistical relation between galaxy and mass is generally complicated due to the non - linear nature of galaxy formation processes , and the linear biasing prescription would not hold even at large scales accessible to the future galaxy surveys . although the extension of the linear biasing prescription to include non - linear effects is rather non - trivial , a straightforward and frequently used prescription for galaxy biasing , coupled to perturbation theory , is the local biasing @xcite , in which the galaxy density field @xmath151 at a given position @xmath152 is described as the local function of mass density field at the same position , i.e. , @xmath153 $ ] . on large scales of our interest ,
this can be expressed as the taylor series expansion : @xmath154=b_1\,\delta({\mbox{\boldmath$x$}}|r ) + \frac{b_2}{2}\,\bigl\{\,[\,\delta({\mbox{\boldmath$x$}}|r)\,]^2 - \bigl\langle\,[\delta({\mbox{\boldmath$x$}}|r)]^2\,\bigr\rangle\,\bigr\ } \nonumber\\ & & \quad\quad\quad\quad + \,\ , \frac{b_3}{3!}\,\bigl\{\,[\,\delta({\mbox{\boldmath$x$}}|r)\,]^3- \bigl\langle\,[\delta({\mbox{\boldmath$x$}}|r)]^3\,\bigr\rangle\,\bigr\ } \,+\,\cdots , \label{eq : local_bias}\end{aligned}\ ] ] where the quantities @xmath155 and @xmath156 are the galaxy and mass density fields smoothed over the radius @xmath157 centered at the position @xmath152 . in the above expression , the coefficients @xmath158 and @xmath159 describe the non - linearity of galaxy biasing , which are incorporated into the galaxy power spectrum @xmath160 , through the perturbative calculation of the matter power spectrum ( @xcite , see also @xcite for different parametrization schemes ) .
although the relation ( [ eq : local_bias ] ) is just a phenomenological prescription , it has been recently applied to the characterization of the baos , and the model of galaxy power spectra has been tested against numerical simulations of halo / galaxy clustering in the case of gaussian initial condition @xcite .
when we consider primordial non - gaussianity , several non - trivial corrections to the galaxy power spectrum @xmath161 appears , which can further alter the shape of the power spectrum on large scales .
this has been first pointed out by mcdonald @xcite . here , we rephrase his finding and advocate its observational importance , together with potential problems .
let us focus on the scales larger than the characteristic scale of baos , above which the higher - order perturbations can be safely neglected .
collecting the relevant terms in the expansion ( [ eq : local_bias ] ) , the power spectrum of the galaxy density fields smoothed over the radius @xmath157 may be expressed as @xmath162 up to the third - order in mass density field , i.e. , @xmath163 . here , @xmath164 is the filter function defined in fourier space . in equation ( [ eq : p_gal_pt ] ) , there are two types of non - gaussian contributions .
one is the term @xmath165 coming from the matter power spectrum , and the other is the function @xmath166 arising from the non - linear mapping of the galaxy biasing .
the explicit expression of the function @xmath167 is @xmath168 the functional form of @xmath167 is clearly different from the perturbative corrections of the matter power spectrum , @xmath46 ( eq.([eq : p12 ] ) or ( [ eq : p_ab^22(k ) ] ) ) .
hence , the presence of this term potentially leads to the scale - dependent galaxy biasing , relative to the matter power spectrum . before continuing
, we note that the quantity @xmath169 manifestly depends on the smoothing radius @xmath157 . in the usual sense
, @xmath169 would not be directly related to the power spectrum derived from the unfiltered observations .
it has been put forward in ref .
@xcite that the unfiltered galaxy power spectrum @xmath170 can be recovered through the simple operation : @xmath171 while the above definition still includes the smoothing radius , we regard equation ( [ eq : def_pg ] ) as a direct observable and evaluate it with the appropriate smoothing radius to see what happens .
below , we will separately treat the galaxy power spectra for the local and equilateral models of primordial non - gaussianity .
, given at @xmath57 in the case of the equilateral model .
in this plot , we do not consider gaussian smoothing .
@xmath81 ( left ) ; @xmath172 ( middle ) ; @xmath173 ( right ) .
[ fig : pt_bias_fnl_equilateral],width=453 ] , given at @xmath57 in the case of the equilateral model . in this plot
, we do not consider gaussian smoothing .
@xmath81 ( left ) ; @xmath172 ( middle ) ; @xmath173 ( right ) .
[ fig : pt_bias_fnl_equilateral],width=453 ] figures [ fig : pt_bias_fnl_local ] plots the ratio of the biasing factor , @xmath174 , plotted against the wavenumber , in the case of local model . here , the biasing factor @xmath175 is defined by @xmath176 the function @xmath177 is similarly defined as above , but with @xmath60 . for illustrative purposes , we specifically set the biasing coefficients to @xmath145 and @xmath178 , and the results for @xmath57 are shown for different values of non - gaussian parameter @xmath31 .
three different panels show the dependence of the smoothing radius , adopting gaussian smoothing , @xmath179 , with radii @xmath180mpc ( left ) , @xmath181mpc ( middle ) , and @xmath182mpc ( right ) .
clearly , the ratios of the biasing factor exhibit a strong scale - dependence , especially at @xmath115mpc@xmath116 .
note that the function @xmath167 logarithmically diverges unless we introduce a large smoothing radius @xmath157 .
accordingly , the strength of the scale - dependence sensitively depends on @xmath157 as well as @xmath31 .
the origin of this scale - dependence can be deduced from a simple manipulation as follows . substituting the bispectrum of the local model ( [ eq : bispectrum_local ] ) into the expression ( [ eq : p_gal_pt ] ) , we take the limit @xmath14 .
the correction @xmath183 becomes negligibly small , and the term including the function @xmath167 is the only dominant contribution to the galaxy power spectrum .
we have @xmath184 where we used the fact that the function @xmath185 defined in equation ( [ eq : m_zeta ] ) asymptotically behaves like @xmath186 in the limit of @xmath14 @xmath187 . here , the quantity @xmath188 represents the rms amplitude of the linear density fluctuation @xmath9 . then , from ( [ eq : def_pg ] ) and ( [ eq : def_bias ] ) , we obtain @xmath189^{1/2}. \label{eq : bk_approx}\ ] ] the expression ( [ eq : bk_approx ] ) contains a term inversely proportional to @xmath185 , which leads to the strong scale - dependence at @xmath14 .
now , we recognize the fact that the scale - dependence of the biasing factor depends not only on @xmath157 and @xmath31 , but also on the ratio of biasing coefficients , @xmath190 , and redshift , @xmath191 .
it is interesting to note that similar kinds of behavior can be found in the halo biasing prescription @xcite .
while the halo biasing describes the clustering properties of virialized objects based on the press - schechter theory , the origin of the scale - dependence is qualitatively the same as that of the local biasing mentioned above . in this sense
, the scale - dependent biasing may be a unique character of the local model of primordial non - gaussianity and it would be an important observable of the large - scale structure .
note , however , that the scale - dependent biasing factor in the local biasing seems problematic because of the logarithmic divergence at @xmath192 , which originates from the rms amplitude @xmath188 in the case of a cdm power spectrum .
further , the biasing factor @xmath175 can be ill - defined for the negative @xmath31 or @xmath158 and it eventually becomes complex at @xmath14 .
mcdonald @xcite proposed a renormalization treatment to remove these difficulties by adding a non - local counter term to the relation ( [ eq : local_bias ] ) .
although the addition of a counter term might be a possible solution , uniqueness of this treatment seems is unclear , and thus the physical reason for the non - local counter term is unclear .
in fact , as shown in next subsection , the scale - dependence of the biasing factor turns out to be model - dependent , and the logarithmic divergence disappears when we consider the equilateral model of primordial non - gaussianity . in figure
[ fig : pt_bias_fnl_equilateral ] , we plot the ratio of the biasing factor , @xmath174 , in the equilateral model . here , instead of showing the dependence of smoothing radius
, we neglect the effect of smoothing ( i.e. , @xmath192 ) and just examine the effect of scale - dependent non - gaussianity by changing @xmath35 to @xmath193 ( left ) , @xmath67 ( middle ) and @xmath194 ( right ) . in marked contrast to the local model
, it turns out that the biasing factors @xmath175 in the equilateral model are almost constant on large scales , and the linear deterministic biasing , @xmath195 , is an excellent approximation to the galaxy power spectrum on large scales .
these features are irrespective of the values of @xmath35 , and can be deduced from the asymptotic behavior of the function @xmath167 . in the @xmath14 limit
, we have @xmath196 .
\end{aligned}\ ] ] recalling the power - law nature of the primordial spectrum @xmath197 , the integral in the above equation converges if @xmath198 and @xmath199 . as a result ,
the function @xmath167 scales as @xmath200 and the non - gaussian contribution becomes negligibly smaller than the linear biasing term , @xmath201 .
therefore , as long as the equilateral model of primordial non - gaussianity is concerned , it is hard to constrain @xmath31 through the scale - dependent biasing factor .
although this conclusion comes from the assumption of the local biasing prescription , this would generally hold for other biasing schemes , including the halo biasing prescription . on the other hand ,
a big difference in the scale - dependent biasing factor between the local and equilateral models implies that the galaxy biasing is very sensitive to the shape of the primordial bispectrum .
an essential reason for having a scale - dependent biasing only for the local - type primordial non - gaussianity is that the amplitude of the primordial bispectrum with squeezed configuration , i.e. , @xmath202 with @xmath203 , is greater than that of the equilateral case ( see eq.([eq : def_f ] ) ) .
the local - type primordial non - gaussianity has fairly strong mode - correlations between small and large fourier modes , and thereby the biasing , as a small - scale phenomenon , affects the power spectrum on very large scales
. this generic feature might be very helpful to discriminate between the types of primordial non - gaussianity .
the detection of scale - dependent biasing may have interesting implication for the inflationary model , according to the single - field consistency relation @xcite .
the consistency relation states that as long as the inflation dynamics is driven by a single scalar field , the primordial bispectrum in squeezed configuration becomes very small , irrespective of the type of non - gaussianity .
thus , single - field inflation generically predicts scale - independent biasing on large scales .
therefore , if the presence of scale - dependent biasing on large scales is observationally verified , this strongly disproves the consistency relation , and the inflation dynamics may not be simply characterized by the single scalar field .
in this paper , we have comprehensively studied the effect of primordial non - gaussianity on the power spectrum of large - scale structure .
using perturbation theory , we calculate the leading - order non - gaussian corrections to the matter power spectrum through the non - linear mode coupling of the gravitational evolution . in the weakly non - linear regime
, the signature of primordial non - gaussianity in the matter power spectrum can be characterized by the primordial bispectrum . adopting two representative models of the bispectrum ( local and equilateral )
, we quantitatively estimate the non - gaussian signals on the matter power spectrum .
the primordial non - gaussianity systematically enhances or suppresses the non - linear growth of the power spectrum amplitude at the @xmath204 1 - 2% level .
we then explore the potential impact on the cosmological parameter estimations from future dedicated surveys for bao measurement .
under the assumption of scale - independent linear biasing , the fisher - matrix analysis reveals that while it is hard to detect the primordial non - gaussianity to the level of current constraints , the inclusion of the non - gaussian parameter @xmath31 significantly degrades the constraints on the primordial spectral index @xmath101 and the running of the index @xmath128 . in this respect ,
the cmb prior information for @xmath101 and @xmath128 as well as the non - gaussian parameter is very crucial . on the other hand ,
determination of the distance scale @xmath205 is rather insensitive to the presence or absence of primordial non - gaussianity , and thus the characteristic scale of baos is a robust standard ruler .
we have also considered the effects of primordial non - gaussianity on the galaxy biasing . in the framework of local galaxy biasing , in which the number density of galaxies is described by a local function of the mass density field
, we found that the non - linear mapping of galaxy biasing can modulate the power spectrum and this sensitively depends on the shape of non - gaussianity . in the local model of primordial non - gaussianity , mode - correlations between large scales and small scales are fairly strong , and the non - gaussian correction induces the strong scale - dependent biasing on large - scales , while the scale - independent linear biasing is preserved to a good accuracy in the equilateral model .
these remarkable properties may be very helpful in discriminating between the types of primordial non - gaussianity , especially in connection with the single - field consistency relation .
however , the biasing factor in the local model exhibits an apparent divergence , suggesting that the local biasing prescription may be incompatible with the local model of primordial non - gaussianity .
indeed , the local biasing scheme is just a phenomenological parametrization and the validity of this prescription itself has not yet been tested enough in the presence of non - gaussianity .
further study using simulations is necessary for quantitative prediction of the non - gaussian signature on galaxy biasing .
we would like to thank patrick mcdonald and shun saito for comments and discussion , and erik reese for a careful reading of the manuscript .
at is supported by a grant - in - aid for scientific research from the japan society for the promotion of science ( jsps ) ( no .
18740132 ) .
tm acknowledges support from the ministry of education , culture , sports , science , and technology , grant - in - aid for scientific research ( c ) ( no .
18540260 , 2006 ) .
kk is supported by stfc and rcuk .
this work was supported in part by grant - in - aid for scientific research on priority areas no .
467 `` probing the dark energy through an extremely wide and deep survey with subaru telescope '' , and jsps core - to - core program `` international research network for dark energy '' . | employing the perturbative treatment of gravitational clustering , we discuss possible effects of primordial non - gaussianity on the matter power spectrum . as gravitational clustering develops , the coupling between different fourier modes of density perturbations becomes important and the primordial non - gaussianity which intrinsically possesses a non - trivial mode - correlation can affect the late - time evolution of the power spectrum .
we quantitatively estimate the non - gaussian effect on power spectrum from the perturbation theory . the potential impact on the cosmological parameter estimation using the power spectrum
are investigated based on the fisher - matrix formalism .
in addition , on the basis of the local biasing prescription , non - gaussian effects on the galaxy power spectrum are considered , showing that the scale - dependent biasing arises from a local - type primordial non - gaussianity . on the other hand
, an equilateral - type non - gaussianity does not induce such scale - dependence because of weaker mode - correlations between small and large fourier modes . |
many physical properties in strongly correlated systems have been studied extensively both experimentally and theoretically .
these are not well described , however , by the local density approximation ( lda ) based on the density functional theory ( dft ) .
lda+u method has been developed beyond lda , @xcite and it works reasonably well for the insulating case near metal - insulator transition in strongly correlated systems .
however , low - energy excitations in metallic phase near metal - insulator transition can not be treated by lda+u method and more sophisticated approaches are needed .
recently the dynamical mean field theory ( dmft ) has been developed and applied to some model systems , which led us to a unified picture of both low- and high - energy excitations in the anomalous metallic phase near metal - insulator transition .
@xcite dmft is based on the mapping of many body systems in bulk onto a single impurity problem , where the off - site coulomb interaction can be neglected , subject to a self - consistency condition . the combination of dmft with lda is applicable to discuss realistic systems .
@xcite the combination of dmft with the gw approximation has been developed to include the off - site coulomb interaction .
@xcite to solve the mapped single impurity problem within dmft , one can use several computational schemes such as the quantum monte carlo method ( qmc ) , @xcite the iterative perturbation theory ( ipt ) , @xcite the non - crossing approximation ( nca ) @xcite and the exact diagonalization ( ed ) .
@xcite while qmc is not applicable in the low temperature limit , ed yields only a discrete number of poles for the density of states .
moreover , the exact calculations such as qmc and ed can only be suitable for simple model hamiltonians . to carry out realistic calculations in realistic materials ,
the approximate ones such as ipt or nca would be more suitable due to its cpu - time , though ipt is not applicable for cases of large coulomb interactions and nca can not yield the fermi liquid behavior at low energies and in low temperature limit .
@xcite the aim of the present paper is to apply the generalized dmft - ipt method for multi - bands on lattices @xcite to the doubly degenerate @xmath0 bands and the triply degenerate @xmath1 bands on a simple cubic lattice and discuss the @xmath3-dependent spectra , the local green s function and the optical conductivity near metal - insulator transition . in sec .
[ sec : form ] we will give the general formulation of the dynamical mean field theory and the optical conductivity . in sec .
[ sec : result ] we will show our numerical results such as @xmath3-dependent spectrum , the imaginary part of the local green s function and the optical conductivity for both @xmath4 and @xmath5 cases . finally we summarize our results in sec .
[ sec : conc ] .
we proceed with the hubbard - type hamiltonian : @xmath6 the first and second terms are the single - particle tight - binding hamiltonian . note that index @xmath7 in the sum of eq .
( [ eqn : lda - ham1 ] ) is running for correlated sites , @xmath8 for orbital indices and @xmath9 for spins .
@xmath2 is the on - site coulomb interaction and the exchange interaction is neglected here .
lattice green s function and local green s function are defined as @xmath10^{-1}_{m{\sigma}m'{\sigma } ' } , \label{eqn : gkw}\\ & & g_{m{\sigma}m'{\sigma}'}({i{\omega}_n})= { \frac{1}{v}}\int d{{\mbox{\boldmath $ { k}$}}}~g_{m{\sigma}m'{\sigma}'}({{\mbox{\boldmath $ { k}$}}},{i{\omega}_n } ) , \label{eqn : gw}\end{aligned}\ ] ] where @xmath11 , @xmath12 and @xmath13 are the single - particle part of the hamiltonian in @xmath3-space , the chemical potential and the unit matrix .
here we neglect the @xmath3-dependence of the self - energy @xmath14 within the framework of dmft .
the chemical potential @xmath12 is determined to satisfy the luttinger s theorem .
@xcite the ipt method was developed by kajueter and kotliar for non - degenerate orbital @xcite and generalized by fujiwara et al . for multi - orbital bands @xcite on arbitrary electron occupation . in the ipt scheme
, we assume the self - energy as @xmath15 where @xmath16 is the second - order self - energy , @xmath17 is the total d - electron occupation number , @xmath18 is the occupation number per each orbital and @xmath19 is the degree of the degeneracy with respect to spins and orbitals .
the coefficients @xmath20 and @xmath21 are determined by requiring the self - energy @xmath22 to be exact in the high - frequency limit @xmath23 , and in the atomic limit @xmath24 @xcite .
@xmath16 is explicitly proportional to @xmath25 and @xmath26 .
the local green s function is calculated in a form @xmath27 where the hamiltonian matrix @xmath11 is diagonalized to be @xmath28 by the unitary matrix @xmath29 .
the @xmath3-integration is carried out by using a generalized tetrahedron method .
@xcite the @xmath30 @xmath3-points are used in the @xmath3-integration within the whole brillouin zone . for a paramagnetic and orbital degenerate system ,
the local green s function becomes diagonal after the @xmath3-integration .
the pad approximation is adopted for analytic continuation of the green s function from the matsubara frequencies @xmath31 to the real @xmath32-axis . from the linear response theory ,
the optical conductivity is given as @xmath33 \label{eqn : opt - definition}\end{aligned}\ ] ] the matrix element of the current operator is defined as @xmath34 from eqs.([eqn : giw - def ] ) and ( [ eqn : opt - definition ] ) , the optical conductivity is then calculated as @xmath35 , \label{eqn : opt - kansei}\end{aligned}\ ] ] where @xmath36 , @xmath37 , @xmath38 , and @xmath39 . the @xmath3-integration in eq .
( [ eqn : opt - kansei ] ) should be carefully carried out .
the @xmath3-integration is able to be carried out simply by the generalized tetrahedron method if the integrand has one pole .
since the two - body green s function @xmath40 has two poles , we should rewrite it to separate into two partial fractions as @xmath41 \nonumber\\ & = & { \frac{1}{2}}\mbox{re}\,\left ( { \frac{g_{\alpha}({{\mbox{\boldmath $ { k}$}}},{\epsilon}_1)-g_{\alpha ' } ^*({{\mbox{\boldmath $ { k}$}}},{\epsilon}_2)}{g_{\alpha}^{-1}({{\mbox{\boldmath $ { k}$}}},{\epsilon}_1)-\{g_{\alpha ' } ^*({{\mbox{\boldmath $ { k}$}}},{\epsilon}_2)\}^{-1 } } } \right .
\nonumber\\ & -&\left.{\frac{g_{\alpha}({{\mbox{\boldmath $ { k}$}}},{\epsilon}_1)-g_{\alpha ' } ( { { \mbox{\boldmath $ { k}$}}},{\epsilon}_2)}{g_{\alpha}^{-1}({{\mbox{\boldmath $ { k}$}}},{\epsilon}_1)-g_{\alpha ' } ^{-1}({{\mbox{\boldmath $ { k}$}}},{\epsilon}_2 ) } } \right ) .
\label{eqn : i - henkei}\end{aligned}\ ] ] in case @xmath42 , the denominator of eq .
( [ eqn : i - henkei ] ) is actually @xmath43-independent and the generalized tetrahedron method can be applied . when @xmath44 , we should treat eq .
( [ eqn : i - henkei ] ) separately for @xmath45 and @xmath46 . when @xmath44 and @xmath45 , eq .
( [ eqn : i - henkei ] ) is then rewritten as @xmath47 , \nonumber\\ \label{eqn : i - kansei1}\\ z({\epsilon } ) & = & \left[1-{\frac{\partial { \sigma}({\omega})}{\partial { \omega}}}\bigg| _ { { \omega}\to { \epsilon}}\right]^{-1 } \ .
\label{eqn : renorm}\end{aligned}\ ] ] when @xmath44 and @xmath48 , eq .
( [ eqn : i - henkei ] ) should be rewritten as @xmath49 \ .
\label{eqn : i - kansei2}\end{aligned}\ ] ] in the above two cases , the resultant @xmath50 can be calculated by using the generalized tetrahedron method .
for cases of @xmath1 bands , the @xmath43-dependent green s function is diagonal and this is also true for @xmath50 .
however , for for @xmath0 bands , the @xmath43-dependent green s function has non - zero off - diagonal elements and also @xmath50 has .
then the denominators of eq .
( [ eqn : i - henkei ] ) in case of @xmath48 for @xmath0 bands are @xmath3-dependent and , therefore , @xmath43-integration is not able to be carried out by using the generalized tetrahedron method . for this case
, @xmath3-integration should be carried out as the sum of the integrand at each @xmath51 .
we have applied the above described calculation scheme to the doubly degenerate @xmath0 bands @xmath52 and the triply degenerate @xmath1 bands @xmath53 .
we consider slater - koster - type tight - binding hamiltonian on a simple cubic lattice .
the effective hopping integrals are assumed only between the nearest neighbor pairs and @xmath54 , @xmath55 and @xmath56 .
the relationship among @xmath57 , @xmath58 and @xmath59 is the scaling properties of bare two - center integrals in the lmto method . here
the half band width of @xmath0 orbitals and @xmath1 orbitals are @xmath60 and @xmath61 , respectively
. the off - diagonal elements of @xmath11 of arbitrary @xmath3 are non zero for the @xmath0 bands .
on the other hand , those for the @xmath1 bands are always zero .
the imaginary part of @xmath3-dependent green s function @xmath62 is shown in fig . [
fig : gkw ] ( shaded regions ) with the energy bands @xmath63 ( full lines ) , for @xmath64 and the electron occupation @xmath65 ( a half - filled case ) .
the energy zeroth is set to be at the chemical potential @xmath12 .
energy bands @xmath63 are shifted by @xmath66 , corresponding to lower and upper hubbard bands . within the fermi liquid theory
, the @xmath3-dependent green s function behaves as @xmath67 near @xmath44 in a metallic case , where @xmath68 is a renormalization factor .
then the width of the coherent peak at the fermi energy is scaled by @xmath69 .
these values become constant within the dmft , since the @xmath3-dependence of the self - energy @xmath14 is @xmath3-independent . in the @xmath0 band case ,
peaks appear at the fermi energy for @xmath70 and @xmath71 , which shows that the system is in a metallic phase . for @xmath70 , those peaks strongly depend on @xmath3 and the spectrum becomes sharper rapidly near the fermi energy . on the contrary , for @xmath71 , those peaks hardly depend on @xmath3 . with increasing coulomb interaction in a metallic phase ( @xmath71 ) ,
coherent peaks appear . at @xmath72 ,
the spectrum splits into upper and lower hubbard bands .
this results from the combination of electronic structure calculation and dmft .
the model calculation of dmft with only assuming the density of states is not able to give the @xmath3-dependent spectra .
it must be noted that , in @xmath1 case , the spectrum shows already the coherent peak hardly depending upon @xmath3 at @xmath70 and it splits into upper and lower hubbard bands at @xmath71 . the imaginary part of the local green s function @xmath73 is shown in fig .
[ fig : greencw - n=0.5 ] , for @xmath64 , with the electron occupation @xmath65 ( a half - filled case ) . the positive and negative energy regions , corresponding to the spectrum of excitation from occupied to unoccupied states ( electron ionization spectrum ) and that from unoccupied to occupied states ( electron affinity spectrum ) . in a half - filled case , the spectrum consists of the upper and lower hubbard bands with the electron - hole symmetry . at the critical region ( here , @xmath74 for @xmath0 bands and @xmath75 for @xmath1 bands )
one observes a sharp coherent peak at @xmath76 . in an insulating case ,
the self energy shows the singular behavior @xmath77 near @xmath78 . on the contrary , in a metallic case
, the self energy behaves as @xmath79 near @xmath78 , which is typical for the fermi liquid . as seen in fig .
[ fig : greencw - n=0.5 ] , the critical @xmath80 ratio is different between @xmath0 bands and @xmath1 bands ; @xmath81 for @xmath0 bands and @xmath82 for @xmath1 bands .
figure [ fig : greencw - n=0.5 ] also shows that in the insulating case , the band width is shrinking more at the same @xmath80 ratio for @xmath1 bands than that is for @xmath0 bands .
this is because that the self - energy is much larger for @xmath1 bands than for @xmath0 bands for the same @xmath80 value due to the degree of the degeneracy with respect to spins and orbitals , @xmath19 .
the second - order self - energy is proportional to @xmath83 and it contributes @xmath84 times larger for @xmath1 bands than for @xmath0 bands .
this fact means that the critical @xmath80 ratio of metal - insulator transition decreases with increasing degeneracy , which contradicts to the conventional understandings that the critical @xmath80 ratio would be the same as in the s - orbital case , @xcite or that the critical @xmath80 ratio would increase with it .
@xcite these contradiction could originate from the hamiltonian on a lattice or the symmetry type of the orbitals .
figure [ fig : greencw - u=0.5 - 1 - 2 ] shows the electron occupation dependence of the imaginary part of the local green s functions for @xmath1 bands with @xmath64 .
( see ref . for @xmath0-band . ) in case of @xmath86 and @xmath87 , one observes the coherent peak at @xmath76 for @xmath70 , which is the critical region of metal - insulator transition .
we also observes the gap for the same @xmath18 s at @xmath76 for @xmath71 and system becomes insulator .
these occupations correspond to the cases of the integer filling of the total d - electron number . at the integer filling case of @xmath88 ( @xmath89:integer ) , the states of @xmath88
is completely filled and the states of @xmath90 is completely empty .
then the gap opens between the ionization and affinity spectra .
with small hole doping @xmath91 , in which electron occupation @xmath92 , the spectrum changes drastically at @xmath70 and @xmath71 .
since the ground state is the mixture of the states of @xmath88 and @xmath93 , a sharp peak appears below the lower hubbard band .
this satellite peak corresponds to the electron ionization of @xmath94 .
the ionization and affinity levels for both @xmath88 and @xmath93 states appear in the spectrum .
these satellite structures of all configurations of @xmath95 appear but their intensity depends on the occupation number . with increasing hole doping @xmath91 ,
the electron ionization spectrum gradually increases . on the contrary ,
the electron affinity spectrum loses the intensity .
when the occupation approaches @xmath93 from above , the electron affinity spectrum @xmath96 disappears . from above discussion
, we have had success of the spectrum calculation which simultaneously shows the effects of multiplet structure and dmft such as the electron ionization and affinity levels of different electron occupations , coherent peaks at the fermi energy in the metallic phase and a gap at an integer filling of electrons for sufficiently large coulomb @xmath2 .
figure [ fig : opt - cond - n=0.5 ] shows the optical conductivity in a half - filled case . in a metallic case , ( @xmath97 for @xmath0 bands , @xmath98 for @xmath1 bands )
three peaks appear : ( 1 ) the drude peak at @xmath44 , ( 2 ) a peak of the width @xmath99 at @xmath100 and ( 3 ) a peak of the width @xmath101 at @xmath102 .
the peak ( 2 ) comes from the transition from the coherent state to the upper hubbard band or that from the lower hubbard band to the coherent state .
the peak ( 3 ) at @xmath2 comes from the transition from the lower hubbard band to the upper hubbard band .
the peaks ( 1 ) and ( 2 ) result from the dmft which includes the coherent state . with increasing coulomb @xmath2 in metallic phase , contributions of the peaks ( 1 ) and ( 2 ) decreases . in the insulating case , ( @xmath72 at @xmath0 bands , @xmath103 at @xmath1 bands ) the peaks ( 1 ) and ( 2 ) disappear and only that of ( 3 ) remains . figure [ fig : opt - cond - n=0.5 ] is consistent with the fact that the critical @xmath80 ratio is different between @xmath0 bands and @xmath1 bands [ @xmath104 for @xmath0 bands and @xmath105 for @xmath1 bands ] .
figures [ fig : opt - cond - t2g - u=1.0 ] and [ fig : opt - cond - t2g - u=2.0 ] show the optical conductivity in cases of arbitrary electron occupations . in case of the integer filling of the total d - electron number , such as @xmath86 and @xmath87 , for @xmath71 , the drude peak disappears and one observes only a peak at @xmath102 , which comes from the transition from the lower hubbard band to the upper hubbard band . with small hole doping @xmath91 , in which electron occupation @xmath92 ( @xmath89=an integer number )
, the drude peak appears and the peak at @xmath102 shrinks .
one also observes a peak which is separated from drude peak at @xmath106 for both @xmath70 and @xmath71 . with increasing hole doping @xmath91 , this peak shifts toward lower energy and becomes sharper and merges into the drude peak , because it comes from the transition from the lower hubbard band to the coherent state . at this range ,
the hight of the drude peak becomes higher and sharper .
also we can see that the intensity at the lower energy side of the peak at @xmath107 decreases simultaneously , and this is because that this side of the peak at @xmath108 were mainly contributed by the transition from the coherent state to the upper hubbard band .
when the occupation approaches @xmath93 from the above , the drude peak disappears again for @xmath71 .
optical conductivity successfully shows the ionization and affinity transition and also metal - insulator transition .
this is , of course , consistent with the results of the spectrum .
in conclusion , we have applied the generalized dmft - ipt method to doubly degenerate @xmath0 bands and triply degenerate @xmath1 bands .
the @xmath3-dependent spectrum shows the resonance behavior near the fermi energy at smaller @xmath2 and the coherent peak near metal - insulator transition .
the spectrum shows that the critical @xmath80 ratio of metal - insulator transition is different between @xmath0 bands and @xmath1 bands and this originates from the degeneracy @xmath19 .
in the arbitrary occupation case , the spectrum shows ionization and affinity levels , coherent peaks at the fermi energy in a metallic phase and a gap at an integer filling of electrons for sufficiently large coulomb @xmath2 .
the optical conductivity gives a spectrum consistent with the above discussions . | we apply the dynamical mean field theory ( dmft ) in the iterative perturbation theory ( ipt ) to doubly degenerate @xmath0 bands and triply degenerate @xmath1 bands on a simple cubic lattice and calculate the spectrum and optical conductivity in arbitrary electron occupation .
the spectrum simultaneously shows the effects of multiplet structure and dmft together with the electron ionization and affinity levels of different electron occupations , coherent peaks at the fermi energy in the metallic phase and a gap at an integer filling of electrons for sufficiently large coulomb @xmath2 .
we also calculate the critical value of the coulomb @xmath2 for degenerate orbitals . |
correlated many - body systems have rich variety of quantum phenomena such as mott insulator , high temperature superconductivity , kondo effect , fractional quantum hall effect and so on @xcite .
interacting boson systems also have many interesting physics such as superfluidity accompanied by the bose - einstein condensation ( bec ) .
a quantum spin system is useful for studying such an interacting boson system due to a mapping between interacting spins- and bosons - systems @xcite .
in fact , the bec in quantum anti - ferromagnet tlcucl@xmath2 has been studied experimentally @xcite and theoretically @xcite . moreover ,
many quantum magnets have been examined so far @xcite .
the density of bosons in quantum magnets is tuned by an applied magnetic field , which induces the bec corresponding to the long - ranged magnetic order .
it can be seen in a magnetization curve .
for example , in the dimerized quantum anti - ferromagnet tlcucl@xmath2 , which has a gap in the spin excitation , the magnetic field has a critical value , @xmath3 , where the gap is destroyed and the magnetization becomes finite .
such a behavior is explained by the bec of triplons @xcite , which are triplet states on a dimer and are associated with hard core bosons . near @xmath4 , the density of triplons
is still dilute and will make some kinds of bound states @xcite . increasing the magnetic field , @xmath5 , the magnetization will become constant above a saturation field , @xmath6 . in @xmath7 , strong repulsion between triplons on a lattice can induce magnetization plateaux , which corresponds to mott insulator phase of bosons @xcite .
the triplon bec mentioned above is accepted as a good starting point of understanding magnetization plateaux . in this paper , we proposed a new concept for magnetization plateaux , which is different from the triplon picture .
such a new concept works on a frustrated two - leg spin ladder ( f-2lsl ) .
calculating magnetization curves of the f-2lsl by the density - matrix renormalization - group ( dmrg ) method , we find three fractional magnetization plateaux due to frustration in the case of strong rung coupling .
we interpret the origin of the plateaux by introducing a quasi - spinon constructed by a singlet and a triplet of a spin pair on a rung . in contrast to the triplon bec picture , the magnetization plateaux correspond to the valence bond solid and the solitonic lattice of the quasi - spinons . as a realistic material containing the f-2lsl , bicu@xmath0po@xmath1 ( bcpo )
has been actively studied @xcite .
et al_. have reported that a field - induced phase can be fit by a solitonic excitation of spinons originating from triplons instead of the triplon bec picture @xcite , indicating a reconstruction of the quasi - particles and field - induced second - order phase transition .
our new concept of quasi - spinons will give a hint for fully understanding the field - induced phases of bcpo as well as other frustrated spin chains that exhibit fractional magnetization plateaux @xcite .
the rest of the paper is organized as follows . in sec .
[ sec : model ] , the hamiltonian of the frustrated two - leg spin ladder will be defined , and a way to calculate the magnetization curves using the dmrg method will be explained . in sec .
[ sec : res ] , numerical results of the magnetization curves , which depend on both the rung couplings and the frustration , will be shown in detail
. we will formulate quasi - spin operators to explicitly derive an effective hamiltonian with the strong frustration and the strong rung couplings in an applied magnetic field .
how to interpret the magnetization plateaux in terms of the effective hamiltonian is also discussed together with schematic pictures .
relation between weak and strong rung - coupling limits will be summarized in a table .
finally , summary and discussions will be given in sec .
[ sec : sum ] .
a model hamiltonian of the f-2lsl with an applied magnetic field @xmath5 is given by @xmath8 with @xmath9 where @xmath10 is the @xmath11 spin operator on the @xmath12 site in the upper ( lower ) chain and its @xmath13 component is @xmath14 .
there are three types of anti - ferromagnetic heisenberg interactions , a nearest - neighbor coupling @xmath15 and a next - nearest - neighbor coupling @xmath16 in the leg direction , and a nearest - neighbor coupling on a rung bond @xmath17 . in the limit of weak rung coupling ,
@xmath18 , this model describes decoupled two frustrated spin chains , while a non - frustrated spin ladder is obtained in the limit of weak frustration , @xmath19 .
consequently , this model bridges between a frustrated spin chain and a non - frustrated spin ladder through @xmath17 and @xmath16 .
the preceding studies on this model have shown that there are two different phases as the ground state , i.e. , the columnar - dimer and rung - singlet phases @xcite . in the columnar - dimer phase ,
the ground state is composed of two degenerated states with spontaneously - broken translational symmetry @xcite , while no degeneracy exists in the ground state of the rung - singlet phase . in the former case ,
spinon as a fermionic quasi - particle helps us to understand the magnetic behavior , although the hard - core bosonic particle , triplon , gives a better explanation in the latter one .
we note that this is one example of reconstruction of quasi - particles caused by the second - order phase transition with spontaneously - broken symmetry .
we calculate the magnetization curves in both the weak- and strong - rung - coupling limits by using the dmrg method with an open boundary condition @xcite .
first , we calculate the minimum energies @xmath20 with fixed magnetizations @xmath21 without a magnetic field . then , the energy difference between @xmath20 and @xmath22 determines a magnitude of a magnetic field , @xmath23 at which the magnetization of the ground state changes from @xmath24 to @xmath25 .
therefore , using @xmath26 and magnetic field @xmath5 , the magnetiztion curve @xmath27 is caluclated by , @xmath28 where the @xmath29 is the heaviside step function .
figures [ fig : mh](a)-(c ) show that magnetization curves for rung couplings @xmath30 , 1 , and 10 , with a fixed frustration @xmath31 in a 72-rung system . in the weak rung - coupling limit , @xmath30 , as shown in fig .
[ fig : mh](a ) , we can find a cusp singularity and a plateau at magnetization ratio @xmath32 , which are also reported in a frustrated chain @xcite . with an applied magnetic field ,
a three - fold degeneracy of triplet excitation breaks and one state of them goes down to the singlet ground state .
when the lowest energy of the triplets reaches down to that of the singlet ground state at @xmath33 , the gap - to - gapless transition occurs and the magnetization becomes finite .
a cusp singularity appears as a result of lifshitz transition of jordan - wigner fermion @xcite .
the strong frustration induces a one - third plateau originated from a reconstructed unit cell with the size three times larger than that of an original unit cell .
freezing spins with a magnetic moment make a long - ranged order in the leg direction together with deconfinement of spins between two legs ( see fig .
[ fig : plateaux](a ) ) .
increasing the rung coupling up to @xmath34 in fig .
[ fig : mh](b ) , the one - third plateau is weakened or disappears , although other plateaux emerge at one - half magnetization @xmath35 and two - third one @xmath36 . moreover , with a strong rung coupling in fig .
[ fig : mh](c ) , we can find again the one - third plateau with some cusp singularities .
the three plateaux at magnetization ratios @xmath32 , 1/2 and 2/3 does not satisfy the condition proposed by oshikawa _ et .
@xcite , which is given by , @xmath37 , with spin @xmath38 , periodicity of lattice @xmath39
( @xmath40 for our hamiltonian ) , magnetization @xmath41 at plateaux , and the set of integers @xmath42 . in accordance with the oshikawa - yamanaka - affleck criteria ,
these are caused by spontaneously - broken symmetries of the plateau states in the present system .
since such a plateau state requires a confirmation with other methods like the numerical calculation and a bosonization approach one by one , these could be non - trivial , although potential plateaux and a possible case of spontaneously - broken symmetries have been discussed in several quantum spin models with similar characteristics @xcite .
the frustration dependence of the magnetization curve with the strong rung coupling @xmath43 is also calculated with @xmath44 , 0.4 , and 0.7 in figs .
[ fig : mh](d ) , ( e ) , and ( f ) , respectively .
it is clear that cusp singularities and plateaux are caused by the frustration controlled by @xmath45 , even though the number of singularities and plateau is different from the weak rung - coupling cases shown in figs . [
fig : mh](a)-(c ) .
the spin freezing and long - ranged order with the one - third plateau state appear at @xmath46 .
in contrast to one - third plateau for weak - rung coupling , the periodicity of the freezing spins must be identical between the upper and lower chains because of the strong - rung coupling ( see fig .
[ fig : plateaux](b ) ) . in order to clarify the origin of the magnetization plateaux
, a quasi - spin transformation @xcite is applied into the original hamiltonian ( [ eq : ham0 ] ) with a strong rung coupling . in this limit ,
the hard - core - bosonic picture is still applicable up to @xmath47 .
however , the picture becomes worse in such a large magnetic field , because the symmetry of the triplets is no longer alive . instead of the symmetry of the triplets , a new su(2 ) symmetry constructed by a singlet and
a triplet appears with a quasi - spin transformation given by , @xmath48 , \nonumber\\ t_{j,{\mathrm}{p}}^z&=&\frac{1}{2}\left[(s_{j,{\mathrm}{u}}^z+s_{j,{\mathrm}{l}}^z)+s_{j,{\mathrm}{u}}^+s_{j,{\mathrm}{l}}^-+s_{j,{\mathrm}{u}}^-s_{j,{\mathrm}{l}}^+\right ] , \label{eq : top1}\end{aligned}\ ] ] and @xmath49,\nonumber\\ t_{j,{\mathrm}{m}}^z&=&\frac{1}{2}\left[-(s_{j,{\mathrm}{u}}^z+s_{j,{\mathrm}{l}}^z)+s_{j,{\mathrm}{u}}^+s_{j,{\mathrm}{l}}^-+s_{j,{\mathrm}{u}}^-s_{j,{\mathrm}{l}}^+\right ] .
\label{eq : top2}\end{aligned}\ ] ] with the transformation , @xmath50 operators obey the spin su(2 ) algebra : @xmath51 and @xmath52=0 $ ] for @xmath53 and @xmath54 .
therefore , the transformation makes us possible to consider another su(2 ) quasi - spin model of two different energy scale discussed below .
the original hamiltonian with no perturbations ( @xmath55 ) is rewritten as , @xmath56 where @xmath57 is number operators of the quasi - spins .
the effective chemical potentials for the quasi - spins are given by @xmath58 and @xmath59 .
with perturbations @xmath60 , we consider magnetic fields @xmath61 .
since the number operator @xmath62 is zero or one , the chemical potential @xmath63 is much greater than @xmath64 .
thus , the energy scales of quasi - spins split off by a strong magnetic field , where the low - energy physics is described by the quasi - spin of ` p ' .
since the magnetization process is obtained by the expectation value @xmath65 as a function of @xmath5 and @xmath66 , we can focus the discussion on the low - energy physics described by the quasi - spin of ` p ' . to neglect the high - energy physics , we consider the effective hamiltonian in a hilbert space projected by @xmath67 with @xmath68 . here
, the projected space has the expectation value of the number operator of the ` m ' quasi - spin @xmath69 . in this approximation ,
the up and down states of the quasi - spin of ` p ' correspond a triplet and a singlet states on a bond , respectively ( see also fig .
2 ) . in the projected hilbert space
, we finally find a frustrated spin chain again , where spins are substituted with the quasi - spins as follows @xcite , @xmath70 with @xmath71 , \\
{ \mathcal}{h}_z^\prime & = & - h^\prime \sum_j t_{j,{\mathrm}{p}}^z,\end{aligned}\ ] ] where the xy- and ising - components of quasi - exchange interactions are denoted by @xmath72 and @xmath73 with @xmath74 , respectively .
see the appendix for more details , by which one can find possible extensions of the theory .
this model has several differences from the original spin model in the weak rung coupling limit ( see table [ tab : comp ] ) .
one is anisotropy of the anti - ferromagnetic spin interactions , where every quasi - exchange interaction is xy - like , @xmath75 .
in addition , we should be careful on the meaning of the magnetization of up and down quasi - spins , because the transformation maps the singlet state to _ down _ and the triplet to _ up _ quasi - spins .
thus , the total magnetization with @xmath76-filling of up spins is not @xmath77 but @xmath76 .
it must be also noted that the quasi - spins couple to the quasi - magnetic field given by , @xmath78 .
this causes the zero expectation value of number of up spins without the real magnetic field @xmath79 , that is , the singlet states pack in rung bonds in the ground state . [ cols="<,^,^",options="header " , ] with the help of the quasi - spin model , we can approach the natures of the magnetization process in the strong rung - coupling limit .
firstly , the magnetization @xmath80 with the strong rung coupling corresponds to @xmath81 in the weak rung - coupling limit .
the gapped ground state in the weak rung - coupling limit is understood by the majumdar - gosh state with spontaneously - broken translational symmetry , where the spin singlet dimers pack over the legs .
in the same manner , the quasi - spin singlet dimers pack over the legs , although the magnetic moment @xmath82 is set on a singlet dimer in fig .
[ fig : plateaux](b ) .
the 1/3 magnetization plateau with the weak rung coupling is equivalent to @xmath83 plateau in the strong rung - coupling limit .
the freezing - spin structure and long - ranged order are also found in the strong rung - coupling limit , where the real spins of the upper and lower chains coherently freeze . additionally , we can find a @xmath84 plateau and some cusps in the strong rung - coupling limit , which correspond to the @xmath85 plateau and cusp singularities with low magnetic fields in the weak rung - coupling limit , respectively .
therefore , the plateaux in the strong rung coupling correspond to @xmath86 and @xmath87 plateaux of the frustrated spin chain , although physical roles are played by the quasi - spins ( see fig . [ fig : plateaux](b ) ) . actually , we confirm that the spins freeze coherently between the upper and lower chains in the @xmath88- and @xmath89-plateau states , and that quasi - spin correlation rapidly decays in the @xmath90-plateau state .
moreover , we can see good correspondence between fig .
[ fig : mh](a ) with the weak rung - coupling and the quarter area at the upper right in fig .
[ fig : mh](c ) with the strong rung - coupling including a cusp singularity . since the filling of the quasi - up - spins with the strong rung coupling corresponds to that in the ground state without a magnetic field in the weak rung coupling limit , the magnetization curve in fig .
[ fig : mh](a ) is folded in half in fig .
[ fig : mh](c ) .
thus , we conclude that the reconstruction of the quasi - particles results in the emergence of the isostructural magnetization curves in two different limits .
in summary , we have studied the magnetization curve of the frustrated spin - ladder system using the dmrg method , and the three magnetization plateaux are found at @xmath32 , @xmath90 , and @xmath89 due to frustration . in the limit of strong - rung coupling , the magnetization near the field - induced phase transition ( @xmath47 ) can be understood by the triplons on rungs as hard - core bosons . on the other hand ,
the plateaux in @xmath91 can be explained by the quasi - spinons , which are reconstructed from the singlet and the triplet of spin pairs on a rung .
the plateau at @xmath35 corresponds to the valence bond solid of the quasi - spinons , while the plateaux at @xmath88 and @xmath89 can be associated with the array of quasi - spinons such as soliton lattice .
this is different from the usual bec picture of triplons .
the magnetization curves around @xmath88 and @xmath89 look like to be folded in half with same characteristics .
this is also naturally explained using the effective model of quasi - spinons with spontaneously - broken symmetry
. our results will be useful for frustrated quantum spin ladder systems such as bcpo .
concerning bcpo , it is noted that an anisotropic interaction , i.e. , dzyaloshinsky - moriya ( dm ) interaction , is reported by the inelastic neutron scattering study @xcite .
the dm interaction also change the usual bec picture of triplons on the magnetization process in the spin ladder system @xcite .
the transition will be modified such as the second order one due to the dm interactions , which lead to linear terms of triplons . when the dm interaction is smaller than the rung coupling , our picture of the quasi - spinons on the magnetization curve will not be changed .
effects of such an interaction on the magnetic plateaux will be clarified in the near future .
we would like to thank m. fujita and o. p. sushkov for valuable discussions .
this work was partly supported by grant - in - aid for scientific research ( grant no.25287094 , no.26103006 ,
no.26108716 , no.26247063 , no.26287079 , no.15h03553 , and no.15k05192 ) , by bilateral program from mext , by mext hpci strategic programs for innovative research ( spire ) ( hp140215 ) and computational materials science initiative ( cmsi ) , and by the inter - university cooperative research program of imr , tohoku university . numerical computation in this work was carried out on the supercomputers at jaea , the k computer at the riken advanced institute for computational science , and the supercomputer center at institute for solid state physics , university of tokyo .
in this section , we present a derivation of effective hamiltonian with a quasi - spin transformation in the strong rung - coupling limit , @xmath92 and @xmath93 . with a finite magnetization in this limit , a quasi - spin transformation and a reduced hamiltonian are useful to understand the ground state and the low - energy physics . in order to obtain the reduced hamiltonian ,
we start with two spin hamiltonian on @xmath12-th rung as follows , @xmath94 with a jordan - wigner transformation of spin operators , @xmath95 and the number operators @xmath96 . as it is well - known , these jordan - wigner operators obey anti - commutation relation , @xmath97 and @xmath98 . to diagonalize this hamiltonian , we can use a bonding and an anti - bonding operators for create and annihilate operators of jordan - wigner fermions , @xmath99 , @xmath100 . with the number operator of the bonding and an anti - bonding jordan - wigner fermions @xmath101 , the rung hamiltonian is rewritten as , @xmath102 we can obtain a quasi - spin transformation eqs .
( [ eq : top1 ] ) and ( [ eq : top2 ] ) with an inverse jordan - wigner transformation as follows , @xmath103 and @xmath104 these operators also satisfy the spin su(2 ) algebra for themselves , and commutate each other .
the quasi - spin operators rewrite the rung hamiltonian as , @xmath105 where effective quasi - magnetic fields , @xmath106 with the quasi - spin operators , the leg hamiltonian between @xmath12-th and @xmath107-th rungs is given by , @xmath108 + t_{j,{\mathrm}{m}}^+ t_{j+l,{\mathrm}{m}}^-\cos\left [ \frac{\pi}{2}(t_{j,{\mathrm}{p}}^z - t_{j+l,{\mathrm}{p}}^z)\right ] \nonumber\\ & & { \hspace{2em}}{\hspace{2em}}- t_{j,{\mathrm}{p}}^+ t_{j+l,{\mathrm}{m}}^+ \cos\left [ \frac{\pi}{2}(t_{j,{\mathrm}{m}}^z - t_{j+l,{\mathrm}{p}}^z)\right ] -t_{j,{\mathrm}{m}}^- t_{j+l,{\mathrm}{p}}^-\cos\left [ \frac{\pi}{2}(t_{j,{\mathrm}{p}}^z - t_{j+l,{\mathrm}{m}}^z)\right ] + { \mathrm}{h .
c. } \bigg\}.\end{aligned}\ ] ] if we consider small quasi - magnetic field for @xmath109 spin , namely @xmath110 , the quasi - magnetic field for @xmath111 spin is much larger than @xmath112 , namely @xmath113 . to deal low - energy physics , we can project out the high - energy states , that is , quasi - up - spins of @xmath111 opeartors . with the projection operator given by @xmath114 ,
an effective hamiltonian is obtained as @xmath115 . since the original hamiltonian ( [ eq : ham0 ] ) is composed by a sum over the rung and leg hamiltonians , @xmath116 and @xmath117 , the quasi - spin transformation gives us the effective hamiltonian eq .
( [ eq : ham1 ] ) .
we note that the quasi - spin operators can be written by singlet and triplets configurations on @xmath12-th rung as follows , @xmath118 and @xmath119 where @xmath120 @xmath121 and @xmath122 denote the triplets and singlet states on @xmath12-th rung , respectively . when the quasi - up - spin states of @xmath111 operators are projected out , the quasi - up - spin and quasi - down - spin of @xmath109 operators approximately correspond to the triplet state @xmath123 and the singlet state @xmath122 , respectively ( see fig . [
fig : top ] ) . -th rung .
when the quasi - up - spin states of @xmath111 operators are projected out , the left - down side is prohibited .
thus , the quasi - up - spin and quasi - down - spin of @xmath109 operators approximately correspond to the triplet state @xmath123 and the singlet state @xmath122 , respectively.,scaledwidth=30.0% ] 00 s. maekawa , t. tohyama , s.e .
barnes , s. ishihara , w. koshibae , g. khaliullin , _ physics of transition metal oxides " _
, springer - verlag berlin ( 2004 ) .
p. fulde , _ correlated electrons in quantum matter " _ , world scientific co. ltd . , singapore ( 2012 ) .
t. matsubara and h. matsuda , prog .
* 16 * , 569 ( 1956 ) .
t. nikuni , m. oshikawa , a. oosawa , and h. tanaka , phys .
lett . * 84 * , 5868 ( 2000 ) .
regg , n. cavadini , a. furrer , h .- u .
gdel , k. krmer , h. mutka , a. wildes , k. habicht , and p. vorderwisch , nature * 423 * , 62 ( 2003 ) .
t. giamarchi and a. m. tsvelik , phys . rev .
b * 59 * , 11398 ( 1999 ) .
t. giamarchi , c. regg , and o. tchernyshyov , nature physics * 4 * , 198 ( 2008 ) .
v. zapf , m. jaime , and c.d .
batista , rev .
* 86 * , 563 ( 2014 ) .
k. hida , j. phys .
. jpn . * 63 * , 2359 ( 1994 ) .
m. oshikawa , m. yamanaka , and i.affleck , phys .
* 78 * , 1984 ( 1997 ) .
k. totsuka , phys .
b * 57 * , 3454 ( 1998 ) .
k. okunishi and t. tonegawa , j. phys .
. jpn . * 72 * , 479 ( 2003 ) ; phys .
b * 68 * , 224422 ( 2003 ) .
a. a. tsirlin , i. rousochatzakis , d. kasinathan , o. janson , r. nath , f. weickert , c. geibel , a. m. lauchli , and h. rosner , phys .
b * 82 * , 144426 ( 2010 ) .
y. kohama , s. wang , a. uchida , k. prsa , s. zvyagin , y. skourski , r. d. mcdonald , l. balicas , h. m. ronnow , c. regg , and m. jaime , phys .
lett . * 109 * , 167204 ( 2012 ) .
f. casola , t. shiroka , a. feiguin , s. wang , m. s. grbi , m. horvati , s. krmer , s. mukhopadhyay , k. conder , c. berthier , h .-
ott , h. m. rnnow , ch .
rugg , and j. mesot , phys .
* 110 * , 187201 ( 2013 ) .
a. lavarlo , g. roux , and n. laflorencie , phys .
b * 84 * , 144407 ( 2011 ) . t. sugimoto , m. mori , t. tohyama , and s. maekawa , phys .
b * 87 * , 155143 ( 2013 ) .
y. kohama , k. mochidzuki , t. terashima , a. miyata , a. demuer , t. klein , c. marcenat , z. l. dun , h. zhou , g. li , l. balicas , n. abe , y. h. matsuda , s. takeyama , a. matsuo , and k. kindo , phys .
b * 90 * , 060408(r ) ( 2014 ) .
t. sugimoto , m. mori , t. tohyama , and s. maekawa , arxiv:1409.4280 . the convergence of energy is confirmed up to a 72-rung system with the truncation number , @xmath124 .
emergence of such a _ non - trivial _ magnetization - plateau state has already been mentioned by k. totsuka with a demonstration of the @xmath125 plateau emerging in the bond - alternating spin-@xmath90 heisenberg chain @xcite .
plumb , k. hwang , y. qiu , l. w. harriger , g. e. granroth , g. j. shu , f.c .
chou , ch .
regg , y. b. kim , and y .- j .
kim , arxive:1408.2528 .
j. sirker , a. weisse , and o.p .
sushkov , europhys .
68 * , 275 ( 2004 ) . | the quantum phase transitions induced by a magnetic field are theoretically studied in a frustrated two - leg spin ladder . using the density - matrix renormalization - group method , we find some magnetic phase transitions and plateaux in two different cases of strong and weak rung couplings . with the strong rung coupling ,
the three magnetization plateaux are found at 1/3 , 1/2 , and 2/3 due to the frustration .
those can be understood in terms of a quasi - spinon reconstructed from the singlet and the triplets of spins on a rung .
the plateau at 1/2 corresponds to the valence bond solid of the quasi - spinons , while the plateaux at 1/3 and 2/3 can be associated with the array of quasi - spinons such as soliton lattice .
this is different from the usual bose - einstein - condensation picture of triplons .
our results will be useful to understand magnetization curves in bicu@xmath0po@xmath1 . |
the first attempt to model the icm in the framework of the hierarchical scenario assumed its thermodynamical properties to be entirely determined by gravitational processes , like adiabatic compression during collapse and shock heating ( kaiser 1986 ) . if no characteristic scales are present in the underlying cosmology ( i.e. , einstein de - sitter cosmology and power law shape for the power spectrum of density perturbations ) , this model should predict hot gas within rich clusters to look the same as within poor groups , since gravity in itself does not have characteristic scales . under the assumptions of emissivity dominated by free free bremsstrahlung and of hydrostatic equilibrium of the gas , this model predicts @xmath1 for the shape and evolution of the relation between @xmath2ray luminosity and icm temperature ( also eke , navarro & frenk 1998 ) .
furthermore , if we define the gas entropy as @xmath3 ( @xmath4 : electron number density ; e.g. , eke et al .
1998 ) , then the self
similar icm has @xmath5 .
however , soon the simple model described above failed to account for several new observational facts : ( a ) the @xmath6@xmath7 relation for nearby clusters is steeper than predicted , with @xmath8 for @xmath9 kev clusters ( e.g. , david et al . 1993 ; white , jones & forman 1997 ; allen & fabian 1998 ; markevitch 1998 ) ; with a possible further steepening at the group scale , @xmath10 kev ( ponman et al .
1996 ; helsdon & ponman 2000 ) ; ( b ) no evidence of evolution for its amplitude has been detected out to @xmath11 ( e.g. , fairley et al . 2000 ; della ceca et al .
2001 ; borgani et al .
2001a ) ; ( c ) the gas density profiles in central regions of cooler groups is relatively softer than for hotter cluster and , correspondingly , the entropy is higher ( e.g. , ponman , cannon & navarro 1999 ; lloyd davis et al .
observational evidence provides a fairly robust measure of the entropy floor for the diffuse hot gas in groups ( lloyd davis et al .
2000 ) of 100 kev/@xmath0 .
a related , very important measurement is the fraction of baryons that has been converted into a `` cold '' , non x ray emitting phase ( i.e stars and neutral hydrogen ) .
this constraint is crucial as it defines the amount of energy available from sne and the importance of cooling in groups and clusters ( renzini 2000 ) .
balogh et al . ( 2001 ) estimate that only a small fraction ( 510% ) of baryons has been converted into stars , almost independently of the total virial mass of the cluster .
@xmath12 _ gas cooling_. at the center of groups the cooling time is much shorter than a hubble time .
cooling would remove low entropy gas transforming it into stars and possibly originate the entropy floor observed .
while advocated on the ground of its simplicity ( muanwong et al 2001 also voit & bryan 2001 ) , the amount of gas involved can not exceed the observed fraction of 515% of the total baryon fraction .
@xmath12 _ energy injection from agns _ ( e.g bhringer , this proceedings ) would be an important source of heating .
most of the energy would come from jets that would carve `` bubbles '' in the icm ( mc namara et al 2001 , quilis et al .
2001 ) , while the x - ray emission from the central engine in radio quiet qsos would likely have a small effect due to the small cross section for the process .
@xmath12 _ energy injection from sne_. winds propelled by sne explosions ( an hefty 10@xmath13erg / sn ) would transfer heat to the icm , possibly escaping the halos of individual galaxies in the forming protocluster and enrich the icm with metals .
( note : substantial metal enrichment could originate also from ram pressure and tidal stripping of galaxies ( moore , quilis & bower 2001 ) .
@xmath12 _ analytical methods _ allow a fast exploration of parameter space , but must sometimes rely on the parametrization of the end results of complicated physical processes . while this approach requires studying the physics of gas in a somewhat idealized situation ( spherical collapse , accretion from a uniform background ) the amount of insight gained is impressive , although conclusions sometimes differ widely ( kaiser 1991,valageas & silk 1999 , tozzi & norman 2000 , bryan 2000 , balogh , babul & patton 1999 , among many ) .
the most refined methods are based on the standard machinery developed to link gravitational clustering in cold dark matter models and the physics of gas and star formation ( cole et al .
2000 , somerville & primack 1999 , bower et al .
2001 , menci & cavaliere 2000 , cavaliere , giacconi & menci 2000 ) .
@xmath12_n - body simulations _ are a powerful method , able to treat the highly non linear problem of the formation of cosmic structures with very few assumptions .
recent observational evidence for a mono phase icm ( bhringer et al 2001 ) remove some potential worries about the use of sph methods , which do not describe well a multiphase medium .
some recent works on the l
t relation including feedback and star formation ( metzler & evrard 1994 , bialek , evrard & mohr 2001 , borgani et al 2001b , valdarnini 2001 ) provide a good starting point and an extensive list of references to the subject of simulations of galaxy clusters .
as mentioned above , several papers suggested that simply the addition of cooling would remove enough gas from the icm to reconcile models with the observed l t relation .
this would likely be in contrast with observational evidence that points to a limited fraction of baryons in clusters in a `` cold '' non x ray emitting phase
. moreover it would be difficult to hide the large quantity of cooled gas in `` dark baryons '' as they would cause detectable deepening of the central potential well at the center of clusters ( e.g lewis et al . 1999 , valdarnini 2001 ) .
this would have a substantial effect on the predicted m
t relation increasing the emission weighted temperature of groups ( finoguenov et al 2001 ) . on the other hand
recent theoretical works , both analytical and numerical seem to ( perhaps slowly ) converge to a required energy input comparable to 100% of what is available from sne ( kravtsov & yepes 2000 , bower et al 2001 , or even a few times higher ( borgani et al . 2001b , valageas & silk 1999 , wu , fabian & nulsen 2000 ) to satisfy the l t relation .
while semi - analytical models estimate the energy available from sne to be of the order of @xmath14 0.25 kev per particle , renzini ( 2000 ) suggested an even smaller energy budget available from sne : just 0.1 kev .
taken together these works suggest that the required energy must be provided by both sne and agns , perhaps active as major mergers shape the early type population of cluster galaxies .
unfortunately none of the n body simulations cited above ( some analytical works did ) treated _ both _ cooling and heating in a fully consistent way or included the effect of agns on the icm .
this leaves the quantitative contribution of cooling and heating rather uncertain .
not really . as the entropy s depends on a inverse power of @xmath15 it might seem more efficient to heat the icm at lower z ( say z@xmath16 ) as suggested in tozzi & norman ( 1999 ) .
however , in their model the authors heated the gas _ before _ accretion , while at the background density . in a more realistic case gas
will be heated when already inside halos that will later merge to form the cluster .
while this requires more energy for a given entropy s , ( @xmath15 is higher by a factor @xmath17 ) this is more than compensated by the fact that halos collapsed at high z will form the core of the cluster at lower z ( governato et al .
2001 ) , i.e. you dump energy right where it is needed .
borgani et al ( 2001b , 2001c ) and bialek , evrard & mohr ( 2000 ) tried a large range of z ( 1 to 5 and 20 respectively ) finding a weak dependence on results for the argument given above .
of course , a redshift too close to z @xmath181 for heating would cause the l t to evolve significantly at moderate redshift , contrary to observations .
bottom line : it is very likely that a combination of cooling & feedback will give the correct , non evolving l
t relation , although only stronger observational constraints and more refined simulations will allow us to evaluate the _ quantitative _ role of cooling , sne and agns .
numerical simulations including a treatment of hydrodynamical processes have reached a high degree of maturity , with widely different codes giving comparable results ( frenk et al 1999 ) . however , the number of particles used to simulate _ individual _ halos is still systematically lower than in dm only runs ( e.g ghigna et al .
the hot debate on the shape of inner density profiles in dm halos has started a race to higher resolution runs ( governato , ghigna & moore 2001 ) . while the quest for the ultimate density profile of dm halos might ultimately prove futile it has sparked a number of improvements on codes and , equally important , increased the robustness of results .
a key method is the so called `` renormalization technique '' ( katz & white 1993 ) where a halo taken from a cosmological simulation is resimulated at much higher resolution , while sparsely resampling the surrounding region .
fig.1 shows the crucial effect of increasing resolution on a sample of sph simulations of renormalized halos .
( borgani et al 2001b , c ) .
t relation obtained from simulations of large cosmological volumes ( where particle number and the ratio softening / r@xmath19 for each halo depends on its virial mass ) likely suffer of systematic biases due to the worsening resolution for less massive halos : namely smaller l@xmath20 and slower cooling due to lower central densities .
0.3truecm this is my very personal wish list for future theoretical and observational studies on the subject : 0.2truecm @xmath12 estimates of l@xmath21 and t in groups that exclude contamination from agns .
@xmath12 estimates of the amount of metals in clusters to a significant fraction of r@xmath19 as another method to infer the energy release from sne ( renzini 2000 , de grandi & molendi 2001 ) .
@xmath12 a robust estimate of the fraction of `` cold '' baryons which keeps into account the amount of intracluster stars .
recent claims ( see arnaboldi et al 2001 ) based on intracluster pns almost double the the efficiency of star formation in clusters .
@xmath12 simulations of sufficient resolution to treat cooling and star formation processes in a robust way .
( note : robust does not mean realistic , but it s a start ) . my checklist for simulating an individual cluster : 0.5truecm n@xmath2210@xmath23 + @xmath24/r@xmath25 + n@xmath26r@xmath27 10@xmath23 + simulation box @xmath14 100mpc + n@xmath28 @xmath29 10@xmath30 + @xmath12 exploring the role of qso s and agns in heating the icm in a cosmological context .
@xmath12 introducing heating schemes that closely follow the time evolution of the physical process behind it .
@xmath12 resolving the discrepancy between adiabatic runs with mesh and sph codes ( e.g kravtsov et al .
2001 , also springel & hernquist 2001 and cen & ostriker 1999 ) : why sph simulations do not have an entropy floor and mesh codes do ?
i wish to thank stefano borgani , marino mezzetti and riccardo valdarnini for organizing such an interesting and lively meeting and uros seljak and bepi tormen for stimulating discussions ( although they made me skip a coffee break or two ) .
i also thank my collaborators stefano borgani , nicola menci , tom quinn , joachim stadel , paolo tozzi & james wadsley for allowing me to discuss our results and show fig.1 . in advance of publication .
allen s.w . , fabian a.c . , 1998 , , 297 , 63 + arnaboldi et al , 2001 aj in press + balogh m.l . , babul a. , patton d.r . , 1999 , , 307 , 463 + balogh et al . , 2001 , , 326 , 1228 + bialek , j.j . , evrard , a.e
mohr , j.j , 2001 , , 555 , 597 + bhringer , h. et al 2001 , a&a , in press + borgani et al .
2001a , 561 , 13 + borgani et al . 2001b , , 559 , 71 + borgani et al .
2001c , in preparation + bower r.g . ,
benson a.j . , bough c.l . ,
cole s. , frenk c.s . , 2001 325 , 497 + bryan g.l . 2000 , , 544 , 1 + cavaliere a. , giacconi r. , menci n. , 2000 , , 528 , l77 + cen r. , & ostriker j.p . , 1999 517
, 31 + cole , s. , lacey , c. g. , baugh , c. m. , & frenk , c. s. 2000 , , 319 , 168 + david l.p .
, slyz a. , jones c. , forman w. , vrtilek s.d . ,
arnaud k.a . , 1993 , , 412 , 479 + de grandi s. & molendi s. , 2001 , 551 , 153 + della ceca r. et al .
2000 , a&a , 353 , 498 + eke , v. , navarro j.f . &
frenk c.s .
199,8 , 503 , 569 + fairley b.w .
et al . , 2000 , , 315 , 669 + finoguenov a. , reiprich t.h . , bhringer h. , 2001 , a&a + frenk , c.s .
et al . 1999 , , 525 , 554 + ghigna s. et al .
2000 , , 544 , 616 + governato , f. , ghigna , s. & moore , b. 2001 , proceedings of astrophysical ages and timescales , ed .
ted von hippel , ( astro - ph/0105443 ) + governato , f. , ghigna , s. , moore , b. , quinn , t. , stadel , j. , & lake , g. 2001 , , 547 , 555 + helsdon s.f .
, ponman t.j . , 2000 , , 315 , 356 + kaiser n. , 1986 , , 222 , 323 + kaiser n. , 1991 , , 383 , 104 + katz , n. & white , s.d.m . , 1993 412 , 455 + kravtsov a.v . , yepes g. , 2000 , , 318 , 227 + kravtsov a.v .
et al .
submitted , ( astro ph/0109077 ) + lewis g.f .
, babul a. , katz n. quinn t. , hernquist l. , 2000 , , 536 , 623 + lloyd - davies e.j . ,
ponman t.j . ,
cannon d.b . , 2000 , , 315 , 689 + markevitch m. , 1998 , apj , 504 , 27 + mc namara , b.r .
et al 2001 , , 562 , 149 + muanwong et al 2001 , 2001 552 , 27 + menci n. , cavaliere a. , 2000 , , 311 , 50 + metzler c.a . ,
evrard , a.e .
1994 437 , 564 + moore b. , quilis v. & bower , r. 2000 , asp conf .
197 , 363 + ponman t.j .
, bourner p.d.j . , ebeling h. , bhringer h. , 1996 , 293 , 690 + ponman t.j . , cannon d.b . , navarro j.f . , 1999 ,
nature , 397 , 135 + quilis , v. , bower , r.g . &
balogh , m.l .
2001 , astro - ph/0109022 + renzini a. , 2000 astro - ph/0001312 , proccedings of lss in the x ray universe , 1999 , ed atlantisscience .
+ somerville , r.s . , & primack , j.r .
1999 , , 310 , 1087 + springel , v. & hernquist l. 2001 astro - ph/0111016 + tozzi p. , norman c. 2001 , , 546 , 63 + valageas p. , silk j. , 1999 , a&a , 350 , 725 + valdarnini r. , 2001 , , in press astro ph/0110545 + voit g.m . , & bryan g.l . , 2001 , nature , in press + white d.a . , jones c. , forman w. , 1997 , , 292 , 419 + wu k.k.s .
, fabian a.c . , nulsen p.e.j . , 2000 , , 318 , 889 + | efforts to understand the deviation of the l
t relation from a simple scaling law valid for clusters and groups have triggered a number of interesting studies on the subject .
techniques and approaches differ widely but most works agree on the important role played by gas cooling and heating sources like agns and sne .
observations set useful constraints on the evolution of the intracluster medium ( icm ) : a 100kev/@xmath0 entropy floor in the core of groups and about 515% of baryons being converted into stars .
however , essential details like the nature of the dominant heating mechanism and the quantitative importance of cooling still need to be addressed .
i suggest that a new generation of high resolution n - body simulations and a quantitative comparison of results between different approaches is required to improve results and increase our understanding of the problem .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
we will prove a general result that will have six seemingly unrelated classes of number theoretic applications .
unfortunately , it will take several pages to state this result .
after this we describe the applications and then prove our theorem . the @xmath0-cantor series expansion , first studied by g. cantor in @xciteis a natural generalization of the @xmath1-ary expansion .
let @xmath2 .
if @xmath3 , then we say that @xmath0 is a * basic sequence*. if @xmath4 , then we say that @xmath0 is * infinite in limit*. given a basic sequence @xmath5 , the * @xmath0-cantor series expansion * of a real @xmath6 in @xmath7 is the ( unique ) expansion of the form @xmath8 where @xmath9 and @xmath10 is in @xmath11 for @xmath12 with @xmath13 infinitely often
. we will write @xmath14 in place of @xmath10 when there is no room for confusion .
moreover , we will abbreviate ( [ eq : cseries ] ) with the notation @xmath15 .
clearly , the @xmath1-ary expansion is a special case of ( [ eq : cseries ] ) where @xmath16 for all @xmath17 .
if one thinks of a @xmath1-ary expansion as representing an outcome of repeatedly rolling a fair @xmath1-sided die , then a @xmath0-cantor series expansion may be thought of as representing an outcome of rolling a fair @xmath18 sided die , followed by a fair @xmath19 sided die and so on .
the study of normal numbers and other statistical properties of real numbers with respect to large classes of cantor series expansions was started by p. erds , a. rnyi and p. turn .
this early work was done by p. erds and a. rnyi in @xcite and @xcite and by a. rnyi in @xcite , @xcite , and @xcite and by p. turn in @xcite .
we recall the following standard definitions ( see @xcite ) .
an * asymptotic distribution function * @xmath20 \to [ 0,1]$ ] is a non - decreasing function such that @xmath21 and @xmath22 . for a sequence of real numbers
@xmath23 with @xmath24 and an interval @xmath25 $ ] , define @xmath26 .
a sequence of real numbers @xmath23 has asymptotic distribution function @xmath27 if @xmath28 for the rest of this paper we will abbreviate asymptotic distribution function as adf .
we say that a sequence @xmath29 is * uniformly distributed mod @xmath30 * if @xmath29 has @xmath31 as its adf . for the rest of this paper
we will abbreviate uniformly distributed mod 1 as u.d .
clearly , not all sequences have an adf .
a sequence of real numbers @xmath23 has * upper asymptotic distribution function * @xmath32 if @xmath33 the sequence @xmath29 has * lower asymptotic distribution function * @xmath34 if @xmath35 every sequence of real numbers @xmath29 has an upper and a lower adf .
we note that the sequence @xmath29 has adf @xmath36 if and only if @xmath37 .
a great deal of information about the @xmath1-ary expansion of a real number @xmath6 may be obtained by studying the distributional properties of the sequence @xmath38 .
for example , it is well known that a real number @xmath6 is normal in base @xmath1 if and only if the sequence @xmath39 is uniformly distributed mod @xmath30 .
thus , we are motivated to make the following definitions for the cantor series expansions . for every basic sequence @xmath0 ,
define @xmath40 and @xmath41 . for integers
@xmath42 and @xmath43 , we define @xmath44 . for any eventually increasing function @xmath45
, we define @xmath46 .
furthermore , set @xmath47 .
the sequences @xmath48 and @xmath49 are defined similarly .
one of the main goals of this paper that will be resolved in main theorem [ thm : main ] will be to show that for a large class of functions @xmath27 the set of real numbers @xmath6 where the sequence @xmath50 has a specified upper and lower adf has full hausdorff dimension .
numerous results will follow from this fact .
we will need to give several definitions to fully describe the scope of this theorem .
we should note that the relationship between the digits of the @xmath0-cantor series expansion of a real number @xmath6 and the sequence @xmath51 is far more complex than the analogous relationship for @xmath1-ary expansions .
the most current results can be found in @xcite .
thus , when generalizing problems involving digits in some @xmath1-ary expansion , we can consider either a problem involving digits in a @xmath0-cantor series expansion or a problem involving the distributional properties of the sequence @xmath51 . often the theory will be different .
for this paper we will always choose the latter option .
the sequence @xmath51 was studied by j. galambos in @xcite and by t. salt in @xcite and several other papers .
the main focus of this paper is to study sets of reals numbers @xmath6 so that @xmath51 and various subsequences of @xmath51 have specific upper and lower adfs .
this will allow us to attack a wide range of problems .
a set of functions @xmath52 is called a * linear family * if for all @xmath42 , @xmath43 , and @xmath53 @xmath54 the set of functions @xmath55 where @xmath56 for all @xmath42 and @xmath43 gives an example of a linear family of adfs .
a non - trivial example is given in the proof of theorem [ thm : normnotap ] in section [ section : otherproofs ] .
for @xmath57 , @xmath58 $ ] , set @xmath59 define a relation @xmath60 among polynomials @xmath57 , @xmath58 $ ] so @xmath61 if @xmath62 and @xmath63 , then we write @xmath64 .
it is easy to verify that @xmath60 is a preorder but not a partial order .
similarly , we can show that @xmath65 is an equivalence relation .
a set of polynomials @xmath66 $ ] is * saturated * if for any @xmath67 $ ] there exists a polynomial @xmath68 and a linear polynomial @xmath69 $ ] such that @xmath70 .
the set @xmath71 is * sparsely intersecting * if for each @xmath72 and @xmath73 we have @xmath74 or @xmath75
. we will prove the following lemma in section [ section : polyisect ] .
[ lemma : iscomplete ] there is a saturated sparsely intersecting set of polynomials . a real number @xmath6 is * computable * if there exists @xmath76 with @xmath77 and a total recursive function @xmath45 that calculates the digits of @xmath6 in base @xmath1 .
a sequence of real numbers @xmath78 is * computable * if there exists a total recursive function @xmath79 such that for all @xmath80 we have that @xmath81 .
a sequence of functions @xmath82 from a metric space @xmath83 to @xmath7 is * uniformly computable * if the double sequence @xmath84 is computable for any computable sequence @xmath85 and if there is a recursive function @xmath86 such that for all @xmath87 and @xmath88 , we have that @xmath89 implies @xmath90 .
a function @xmath27 is * uniformly computable * if the sequence @xmath91 is uniformly computable .
a sequence @xmath78 is * uniformly computable * if there is a uniformly computable function @xmath92 such that @xmath93 @xcite .
a basic sequence @xmath94 is a * computably growing basic sequence * if it is infinite in limit and the sequence @xmath95 is computable .
a linear family of adfs @xmath52 is an * explicit linear family of adfs * if for each @xmath96 with @xmath97 the following hold . 1 .
the real numbers @xmath98 and @xmath99 : f_{m , r}(x ) = q \right\}}$ ] are computable for every rational @xmath100 $ ] .
if @xmath101 is discontinuous at @xmath102 , then @xmath102 is a computable real number and @xmath103 is a computable real number .
the function @xmath101 is either continuous , has only finitely many discontinuities , or the set of its discontinuities may be written in the form @xmath104 , where @xmath105 is a uniformly computable sequence . a sparsely intersecting set of polynomials @xmath106 is an * explicit sparsely intersecting set of polynomials * if for all @xmath107 there exists a computable sequence @xmath108 such that if @xmath109 , then @xmath110 or @xmath111 . given a basic sequence @xmath0 , a set of sparseley intersecting polynomials @xmath71 , and a set of linear families @xmath112 of adfs , define @xmath113 we may now state the main theorem of this paper .
[ thm : main ] if @xmath0 is infinite in limit , @xmath71 is a set of sparsely intersecting polynomials , and @xmath114 is a set of linear families of upper and lower adfs given by ( [ eq : fadf ] ) , then @xmath115 furthermore , if @xmath0 is computable and computably growing , @xmath71 is explicit , and @xmath114 is explicit , then there is a subset @xmath116 of @xmath117 such that the following hold . 1
. the set @xmath116 has full hausdorff dimension .
2 . there exist computable sequences @xmath118 and @xmath119 such that @xmath120 the set @xmath116 and computable sequences @xmath118 and @xmath119 are constructed in section [ section : construction ] .
main theorem [ thm : main ] is proven in section [ section : construction ] .
we recall the modern definition of a normal number .
[ definition : normal ] a real number @xmath6 is * normal of order @xmath121 in base @xmath1 * if all blocks of digits of length @xmath121 in base @xmath1 occur with relative frequency @xmath122 in the @xmath1-ary expansion of @xmath6 .
moreover , @xmath6 is * simply normal in base @xmath1 * if it is normal of order @xmath30 in base @xmath1 and @xmath6 is * normal in base @xmath1 * if it is normal of order @xmath121 in base @xmath1 for all natural numbers @xmath121 .
it is well known that .
borel @xcite was the first mathematician to study normal numbers . in 1909
he gave the following definition .
[ definition : normalborel ] a real number @xmath6 is * normal in base @xmath1 * if each of the numbers @xmath123 is simply normal ( in the sense of definition [ definition : normal ] ) , in each of the bases @xmath124 .
e. borel proved that lebesgue almost every real number is normal , in the sense of definition [ definition : normalborel ] , in all bases . in 1940 , s. s. pillai @xcite simplified definition [ definition : normalborel ] by proving that [ thm : pillai ] for @xmath125 , a real number @xmath6 is normal in base @xmath1 if and only if it is simply normal in each of the bases @xmath124 .
theorem [ thm : pillai ] was improved in 1951 by i. niven and h. s. zuckerman @xcite who proved [ thm : nivenzuckerman ] definition [ definition : normal ] and definition [ definition : normalborel ] are equivalent .
a simpler proof of theorem [ thm : pillai ] was given by j. e. maxfield in @xcite .
j. w. s. cassels gave a shorter proof of theorem [ thm : nivenzuckerman ] in @xcite .
it should be noted that both of these results require some work to establish , but were assumed without proof by several authors .
for example , m. w. sierpinski assumed theorem [ thm : pillai ] in @xcite without proof .
moreover , d. g. champernowne @xcite , a. h. copeland and p. erds @xcite , and other authors took definition [ definition : normal ] as the definition of a normal number before it was proven that definition [ definition : normal ] and definition [ definition : normalborel ] are equivalent .
more information can be found in chapter 4 of the book of y. bugeaud @xcite .
the following theorem was proven by h. furstenberg in his seminal paper `` disjointness in ergodic theory , minimal sets , and a problem in diophantine approximation '' @xcite on page 23 as an application of disjointness to stochastic sequences .
[ thm : basebnormalii ] suppose that @xmath126 is the @xmath1-ary expansion of @xmath6 .
then @xmath6 is normal in base @xmath1 if and only if for all natural numbers @xmath42 and @xmath43 the real number @xmath127 is normal in base @xmath1 .
we will say that @xmath6 is * ap normal of type i in base @xmath1 * if @xmath6 satisfies definition [ definition : normalborel ] and * ap normal of type ii in base @xmath1 * if @xmath6 satisfies the notion introduced in theorem [ thm : basebnormalii ] .
it is interesting to note that although furstenberg did not provide an alternate proof of theorem [ thm : nivenzuckerman ] , he did define an equivalent notion of normality , that is ap normality of type ii .
see appendix for more details .
thus , for numbers expressed in base @xmath1 @xmath128 the authors feel that the equivalence of definition [ definition : normal ] and definition [ definition : normalborel ] and other similar ones is a far more delicate topic than is typically assumed .
the core of e. borel s definition is that a number is normal in base @xmath1 if blocks of digits occur with the desired relative frequency along all infinite arithmetic progressions .
we say that a real number @xmath6 is * @xmath0-distribution normal * if @xmath51 is u.d .
a real number @xmath6 is * ap @xmath0-distribution normal of type i * if for all @xmath129 and @xmath130 we have that @xmath131 is u.d .
if @xmath132 , then we say that @xmath6 is * ap @xmath0-distribution normal of type ii * if the real number @xmath133 is @xmath134-distribution normal for all @xmath129 and @xmath130 .
we say that @xmath6 is * ap @xmath0-distribution abnormal * if @xmath131 is not u.d .
mod 1 for any @xmath135 .
we will prove in section [ section : otherproofs ] that @xmath0-distribution normality is not equivalent to ap @xmath0-distribution normality in a particularly strong way .
the following theorem describes exactly how much ( [ eq : apequiv ] ) may be extended to @xmath0-cantor series expansions when @xmath0 is infinite in limit .
[ thm : normnotap ] let @xmath0 be a basic sequence that is infinite in limit
. then 1 .
ap @xmath0-distribution normality of type i is equivalent to ap @xmath0-distribution normality of type ii .
the set of real numbers which are @xmath0-distribution normal and ap @xmath0-distribution abnormal is a meagre set with zero measure and full hausdorff dimension .
furthermore , if @xmath0 is computable and computably growing , then the proof of theorem [ thm : normnotap ] provides a computable example of a real number that is @xmath0-distribution normal and ap @xmath0-distribution abnormal .
see section [ section : constructionnormal ] for further discussion .
however , main theorem [ thm : main ] is far stronger since it allows us to specify upper and lower adfs along polynomially indexed subsequences of @xmath51 .
theorem [ thm : normnotap ] only requires knowledge of @xmath51 along infinite arithmetic progressions .
we note that far less is known if we extend ( [ eq : apequiv ] ) to analogous definitions involving digits .
the problem is discussed in @xcite and partial results are given .
one substantial difference is that the analogous generalizations of the definitions of ap normality of types i and ii are no longer equivalent .
however , these definitions are technical , so we choose not to state any of these results here .
the following well known result was proved for @xmath136 by a. s. besicovitch in @xcite and for all other @xmath1 by h. eggleston in @xcite .
[ thm : eggleston ] let @xmath137 and @xmath138 be a probability vector .
then the hausdorff dimension of the set of all real numbers @xmath6 where the digit @xmath72 occurs in the @xmath1-ary expansion of @xmath6 with relative frequency @xmath139 for all @xmath140 is equal to @xmath141 there have been numerous improvements of theorem [ thm : eggleston ] .
moreover , theorem [ thm : eggleston ] has been extended to certain classes of cantor series expansions .
early work was done by j. peyrire in @xcite and y. kifer in @xcite .
we mention a similar result proven by y. xiong in @xcite . for a given basic sequence @xmath0 ,
let @xmath142 denote the number of times a block @xmath143 occurs starting at a position no greater than @xmath17 in the @xmath0-cantor series expansion of @xmath6 .
[ thm : xiong ] suppose that @xmath0 is infinite in limit and that @xmath144 is an infinite probability vector .
for @xmath145 , let @xmath146 then the following hold . 1 .
if @xmath147 , then @xmath148 where @xmath149 2 . if @xmath0 is increasing and the sequence @xmath150 is bounded , then @xmath151 c. m. colebrook @xcite proved a similar result to theorem [ thm : eggleston ] about the hausdorff dimension of the set of real numbers @xmath6 where the sequence @xmath39 has a given adf .
a special case of main theorem [ thm : main ] extends c. m. colebrook s result to a large class of cantor series expansions .
[ thm : extendcolebrook ] suppose that @xmath0 is infinite in limit and that @xmath27 is an adf .
then the set of real numbers @xmath6 such that @xmath51 has adf of @xmath27 has full hausdorff dimension .
we note that the sets considered in theorem [ thm : xiong ] have much smaller hausdorff dimension than those considered in theorem [ thm : extendcolebrook ] .
this is in sharp contrast to the case of the @xmath1-ary expansions .
it is difficult in general to analyze sets of real numbers some of whose digital frequencies may not exist .
this is discussed in l. olsen s paper @xcite . for an integer @xmath125 ,
let @xmath152 denote the number of times a block @xmath143 occurs starting at a position no greater than @xmath17 in the @xmath1-ary expansion of @xmath6 and set @xmath153 [ thm : olsen ] for @xmath154 and @xmath137 , put @xmath155 : \pi_b(1,x)=k\pi_b(0,x ) \right\}};\\ & \mathscr{b}={\left\ { x \in [ 0,1 ] : \pi_2(000,x)=\pi_2(11,x ) \right\}};\\ & \mathscr{c}={\left\ { x \in [ 0,1 ] : \pi_3(1,x)^2=\pi_3(0,x ) \right\}}.\end{aligned}\ ] ] then @xmath156 we note that l. olsen provides a precise description of the hausdorff dimension of @xmath157 and @xmath158 .
we intentionally omit these descriptions here and only provide an approximation as they are relatively complex .
we will prove the following extension of theorem [ thm : olsen ] in section [ section : otherproofs ] .
[ thm : olsengen ] suppose that @xmath159 and @xmath160 are disjoint non - degenerate intervals and that @xmath27 and @xmath161 are real valued functions on @xmath162 $ ] which satisfy the following conditions : * there is an @xmath163 in @xmath164 $ ] such that @xmath165 ; * we have that @xmath166 .
if @xmath0 is infinite in limit , then @xmath167 s. albeverio , m. pratsiovytyi , and g. torbin proved the following theorem in @xcite [ thm : nofrequencyapt ] the set of real numbers whose frequencies of digits in base @xmath1 do not exist has zero measure and full hausdorff dimension .
we extend their result to the following theorem .
[ thm : nofrequency ] if @xmath0 is infinite in limit , then the set of real numbers @xmath6 such that @xmath51 has no adf has zero measure and full hausdorff dimension .
theorem [ thm : nofrequency ] is proven in section [ section : otherproofs ] .
[ section : constructionnormal ] the most well known construction of a normal number in base @xmath168 is due to champernowne .
the number @xmath169 formed by concatenating the digits of every natural number written in increasing order in base 10 , is normal in base 10 .
the number formed by concatenating the digits of the natural numbers in base @xmath1 in order is normal in base @xmath1 .
a. h. copeland and p. erds in @xcite showed that the number @xmath170 formed by concatenating the digits of all prime numbers is normal in base @xmath1 .
h. davenport and p. erds in @xcite showed that the number formed by concatenating the value of a positive integer valued polynomial at each natural number yields a normal number in base @xmath1 .
many similar and more sophisticated results have been proven since then .
for example , j. vandehey @xcite and m. madritsch and r. tichy @xcite have given similar constructions .
a more extensive list of results can be found in y. bugeaud s book @xcite .
a real number is * absolutely normal * if it is normal in base @xmath1 for all @xmath137 .
m. w. sierpiski gave an example of an absolutely normal number that is not computable in @xcite .
the authors feel that examples such as m. w. sierpiski s are not fully explicit since they are not computable real numbers , unlike champernowne s number .
a. m. turing gave the first example of a computable absolutely normal number in an unpublished manuscript .
this paper may be found in his collected works @xcite .
the @xmath17th digit of a. m. turing s number may be computer with an algorithm that is doubly exponential in @xmath17 .
v. becher , p. a. heiber , and t. a. slaman constructed an absolutely normal number in @xcite whose digits may be computed in polynomial time .
see @xcite by v. becher , s. figueira , and r. picchi for further discussion .
we will use main theorem [ thm : main ] to construct a computable @xmath0-distribution normal number when @xmath0 is computable and computably growing .
[ thm : computableelement1 ] suppose that @xmath0 is computable and computably growing , @xmath171 , and @xmath172 .
if @xmath118 is the sequence given in main theorem [ thm : main ] , then the real number @xmath173 is computable and @xmath0-distribution normal .
lafer @xcite asked for a construction of a @xmath0-distribution normal number for an arbitrary basic sequence @xmath0 .
theorem [ thm : computableelement1 ] provides a partial answer to this question .
however , we remark that there are basic sequences @xmath0 such that no @xmath0-distribution normal number is computable .
a. a. beros and k. a. beros showed in @xcite that there exists a limit computable basic sequence @xmath0 such that no computable real number is @xmath0-distribution normal .
thus , it is impossible to answer p. laffer s question for an arbitrary basic sequence .
it remains open how much one needs to assume about a basic sequence @xmath0 in order to guarantee existence of computable @xmath0-distribution normal numbers and how to construct one of these numbers .
there is substantial literature in pertaining to the explicit construction of numbers with different digital frequencies .
the notes of section 1.8 in @xcite provide a good list of papers on the subject .
we solve an analogous problem for cantor series expansions with @xmath0 infinite in limit : constructing a computable real number @xmath6 so that @xmath51 has a given adf @xmath174 .
this follows immediately from main theorem [ thm : main ] and is a more general version of theorem [ thm : computableelement1 ] .
[ thm : computableelement2 ] suppose that the singleton @xmath175 is an explicit set of adfs , @xmath0 is computable and computably growing , @xmath171 , and @xmath176 .
if @xmath118 is the sequence given in main theorem [ thm : main ] , then the real number @xmath177 is computable and @xmath178 has adf @xmath174 .
we mention four results from other papers that will follow as corollaries of our main theorem .
in fact the immediate corollaries will be stronger than the results stated in this section .
the following theorem was proven by j. peyrire in @xcite .
[ thm : pey ] suppose that @xmath0 is infinite in limit .
then for all @xmath179 @xmath180 for any sequence @xmath181 of real numbers , let @xmath182 denote the set of accumulation points of @xmath83 . given a set @xmath183 $ ]
, let @xmath184 the following results are proven by y. wang , z. wen , and l. xi in @xcite .
[ thm : wangwenxi ] if @xmath0 is infinite in limit , then @xmath185 for every closed set @xmath186 .
let @xmath187 note that theorem [ thm : pey ] follows as a direct consequence of theorem [ thm : wangwenxi ] by considering the closed set @xmath188 for @xmath189 .
we will prove the following theorem as a corollary of main theorem [ thm : main ] in section [ section : arith ] .
[ thm : strongerwangwenxi ] suppose that @xmath190 and closed sets @xmath191 are contained in @xmath162 $ ] for @xmath192 , and @xmath97 . if @xmath193 , then @xmath194 note that theorem [ thm : wangwenxi ] follows immediately from theorem [ thm : strongerwangwenxi ] by setting @xmath195 .
similarly , we may obtain theorem [ thm : pey ] by setting @xmath195 and @xmath196 . in @xcite g. myerson and a.
d. pollington proved the following result .
[ thm : mp ] there exists a uniformly distributed sequence @xmath181 such that none of the sequences @xmath197 are u.d .
mod 1 for any @xmath198 .
moreover , g. myerson and a. d. pollington provide an example of a sequence with the property described in theorem [ thm : mp ] .
+ let @xmath199 be given by @xmath200 .
by theorem [ thm : normnotap ] there exists a computable real number @xmath201 that is @xmath0-distribution normal and ap @xmath0-distribution abnormal . setting @xmath202
, we see that theorem [ thm : mp ] follows immediately from theorem [ thm : normnotap ] .
moreover , the sequence @xmath78 is computable .
thus , main theorem [ thm : main ] provides a strong generalization of theorem [ thm : mp ] .
lastly , given any adf @xmath27 , j. g. van der corput constructed a sequence with adf @xmath27 in @xcite .
the proof of main theorem [ thm : main ] gives a far more general construction .
in order to prove main theorem [ thm : main ] , we will need to construct the set @xmath116 .
[ section : polyisect ] to construct a sequence @xmath51 with desired distributional properties along polynomially indexed subsequences we will need to estimate @xmath203 .
we say an equation @xmath204 with @xmath205 $ ] has infinitely many solutions with bounded denominator if there is some @xmath206 such that there are infinitely many @xmath207 with @xmath208 and @xmath204 .
if there are infinitely many integer solutions , there are infinitely many solutions of bounded denominator .
we will need the following theorem by y. bilu and r. tichy @xcite .
[ thm : bilutichy ] let @xmath209 $ ] .
the equation @xmath210 has infinitely many solutions of bounded denominator if and only if there exist a polynomial @xmath211 $ ] , linear polynomials @xmath212 $ ] and polynomials @xmath213 $ ] where @xmath214 is a standard pair such that @xmath215 . the standard pairs are 1 .
@xmath216 with @xmath217 and @xmath218 ; 2 .
@xmath219 ; 3 .
@xmath220 , where @xmath221 is the @xmath42-th dickson polynomial and @xmath222 ; 4 .
@xmath223 with @xmath224 ; 5 .
. where @xmath226 $ ] is non - zero ( possibly constant polynomial , and @xmath227 are non - zero . with this theorem
we can prove the following .
[ lemma : sparseintersect ] if @xmath228 $ ] , then @xmath64 if and only if there exist linear polynomials @xmath229 , @xmath230 @xmath231 $ ] such that @xmath232 . for
the forward direction , suppose that two linear polynomials @xmath233 and @xmath234 exist so that @xmath232
. then @xmath235 similarly , @xmath236 .
for the reverse direction we look at two cases : when @xmath237 and when @xmath238 .
suppose first that @xmath239 .
write @xmath240 then @xmath241 but @xmath242 so @xmath243 similarly , @xmath244 since @xmath245 we have that @xmath246 but @xmath57 and @xmath247 have different degrees , so they can not be equal when composed with linear polynomials . thus the theorem holds for this case .
now suppose that @xmath248 . in order to have that @xmath64
, there must be infinitely many integer solutions to the equation @xmath210 . by theorem [ thm : bilutichy ]
, we must have that there exists a polynomial @xmath211 $ ] , linear polynomials @xmath249 $ ] , and a standard pair of polynomials @xmath213 $ ] . since @xmath237
, we must have that @xmath250 .
but there are only a few cases where a standard pair of polynomials can have the same degree . for standard pairs of the first and third type
, we must have that @xmath27 and @xmath161 are linear .
for standard pairs of the second and fourth type , we must have that @xmath27 and @xmath161 are of the form @xmath251 with @xmath227 .
standard pairs of the fifth kind can not have equal degrees .
if @xmath27 and @xmath161 are linear polynomials , then the proof of the previous direction suffices .
we only need to prove the claim when @xmath27 and @xmath161 are quadratic with zero linear term , or equivalently when @xmath252 and @xmath253 . in this case , note that the equation @xmath254 is equivalent to @xmath255 with integers @xmath256 , @xmath1 , and @xmath257 .
the solutions of this equation are distributed accoring to a number of different relations between @xmath256 , @xmath1 , and @xmath257 .
the only scenario where this equation has infinitely many solutions is when -ab is not square and positive . in that case , the equation can be rewritten as @xmath258 , with @xmath259 , @xmath260 and @xmath261 .
integer solutions to this equation give an upper bound to solutions of the original equation .
this is the generalized pell equation , whose solutions are well known .
they are of the form @xmath262 for @xmath263 where there are finitely many base solutions @xmath264 , and all further solutions are generated by multiplying by units @xmath265 in @xmath266 $ ] .
thus @xmath267 so @xmath268 with this lemma we can now construct a saturated sparsely intersecting set of polynomials .
@xmath269 start by ordering @xmath270 $ ] as follows .
first , list all polynomials of degree less than or equal to @xmath30 with coefficients whose absolute values are less than or equal to 1 .
then list all polynomials of degree at most @xmath271 and coefficients with absolute values at most @xmath271 ordered lexicographically , removing any repeated polynomials . at step @xmath17 ,
list all polynomials of degree at most @xmath17 and coefficients with absolute values at most @xmath17 ordered lexicographically . in this way
, we create a bijection between the natural numbers and @xmath270 $ ] .
if @xmath272 for any @xmath273 , then there exist @xmath247 and @xmath229 , @xmath230 such that @xmath274 and @xmath275 .
replace @xmath276 by @xmath247 , remove any other instances of @xmath247 in the ordering , and remove @xmath139 .
let @xmath277 be the result of this operation completed for @xmath278 .
then @xmath279 is our desired indexed set . in order to construct computable members of @xmath116
we will need to better understand explicit families of adfs and polynomials .
[ lemma : approxadf ] if @xmath280 is an explicit linear family of adfs , then there exists a sequence of explicit linear families of continuous adfs @xmath281 so that @xmath282 converges to @xmath101 pointwise .
we mimic the proof in @xcite that for any adf @xmath27 , there exists a sequence of continuous adfs converging to @xmath27 .
define increasing sequences @xmath283 such that @xmath284 contains @xmath285 for @xmath286 and all @xmath102 so that @xmath287 there are finitely many @xmath102 that satisfy ( [ eq : overnjump ] ) , so @xmath284 is a finite sequence .
each element of @xmath284 is a computable real number as well .
note that @xmath288 .
let @xmath289 and piecewise linear between @xmath290 and @xmath291 .
then @xmath282 is continuous , non - decreasing , @xmath292 , and @xmath293 .
note that for any @xmath294 , since @xmath55 is a linear family of adfs , we have that @xmath295 for all @xmath53 . as @xmath282 is piecewise linear , we have that this equality holds for all @xmath296 $ ] .
so @xmath297 is a linear family of adfs .
as @xmath282 is continuous , we only need to check that @xmath298 and @xmath299 are computable real numbers for all @xmath300 .
we have that @xmath301 for some @xmath72 , so @xmath302 since @xmath282 is piecewise linear .
but the set of computable real numbers is closed under the usual field operations of the reals ( see @xcite ) , so the real number @xmath298 is computable .
if @xmath303 consists of a single point , say @xmath43 , then we have that @xmath304 and @xmath305 which is a computable real number .
if @xmath303 does not consist of a single point , we have that there are maximum and minimum integers @xmath72 and @xmath73 such that @xmath306 . since @xmath307 ,
for any @xmath308 we have that @xmath309 .
so the real number @xmath310 is computable .
thus @xmath282 is an explicit linear family of adfs . to see that @xmath282 converges pointwise to @xmath101 ,
let @xmath311 $ ] . if @xmath102 is a discontinuity of @xmath101 , then @xmath312 , which implies that for some @xmath313 we have that @xmath314 . then @xmath315 for @xmath316 .
now suppose @xmath102 is not a discontinuity of @xmath101 .
let @xmath317 . then for some @xmath313 , if @xmath318 , then @xmath319 .
for some @xmath72 , we have that @xmath320 . as @xmath321 , we know that @xmath322 . thus @xmath323 and @xmath324 .
but since @xmath325 for all @xmath72 and @xmath326 , then we have that @xmath327 .
thus @xmath282 converges pointwise to @xmath101 .
the following result by s. tengely @xcite is useful for constructing an explicit set of polynomials and proving main theorem [ thm : main ] .
[ thm : tengely ] let @xmath209 $ ] be monic polynomials with @xmath328 such that @xmath329 is irreducible in @xmath330 $ ] and @xmath331 .
let @xmath332 be a divisor of @xmath333 . if @xmath334 is a solution of the diophantine equation @xmath210 , then @xmath335 where @xmath336 and @xmath337 denotes the maximum of the absolute values of the coefficients .
we prove the following theorem to show that the second part of main theorem [ thm : main ] is not vacuous .
there is a sparsely intersecting explicit set of polynomials that contains @xmath338 .
we proceed with the construction in the proof of lemma [ lemma : iscomplete ] exactly as before , creating an ordering of all polynomials .
note that by construction , @xmath339 is the identity polynomial .
as before , at step @xmath340 we check if @xmath139 and @xmath276 satisfy the properties of theorem [ thm : tengely ] for @xmath341 .
if they do , then there is a computable bound @xmath342 on the absolute values of solutions to @xmath343 .
thus if we set @xmath344 , we have that if @xmath109 , then @xmath345 since @xmath346 .
this ensures sparse intersection and explicitness when we do not consider @xmath339 .
suppose @xmath347 and @xmath348 .
writing @xmath349 , set @xmath350 .
note that if @xmath351 , then @xmath352 .
thus @xmath353 for @xmath354 .
set @xmath355 .
therefore @xmath356 is a computable sequence .
moreover , if @xmath357 , then @xmath358 .
if @xmath139 and @xmath276 do not satisfy the conditions of theorem [ thm : tengely ] for any @xmath359 , we remove @xmath139 from @xmath71 and relabel @xmath360 to @xmath139 and so on .
let @xmath277 be the resulting set of this procedure conducted for @xmath361 .
set @xmath279 .
we will construct a subset @xmath116 of @xmath117 so that @xmath116 has the structure of a homogeneous moran set and full hausdorff dimension .
let @xmath362 be a sequence of positive integers and @xmath363 be a sequence of positive numbers such that @xmath364 , @xmath365 , @xmath366 , and @xmath367 , where @xmath53 is a positive real number .
for any @xmath121 , let @xmath368 , and @xmath369 , where @xmath370 . if @xmath371 , @xmath372 , put @xmath373 .
suppose @xmath160 is a closed and bounded interval .
the collection of closed subintervals @xmath374 of @xmath160 has * homogeneous moran structure * if : 1 .
@xmath375 ; 2 .
@xmath376 are subintervals of @xmath377 and @xmath378 for @xmath379 ; 3 . @xmath380 , @xmath381 .
suppose that @xmath382 is a collection of closed subintervals of @xmath160 having homogeneous moran structure .
let @xmath383 .
we say @xmath384 is a * homogeneous moran set determined by * @xmath382 , or it is a * homogeneous moran set determined by * @xmath160 , @xmath362 , @xmath385 .
we will need the following theorem of d. feng , z. wen , and j. wu from @xcite .
if @xmath386 is a homogeneous moran set determined by @xmath160 , @xmath387 , @xmath385 , then @xmath388 [ section : construction ] the construction given in this section will be related to the constructions given by the second author in @xcite .
suppose that we are given a computable and computably growing basic sequence @xmath94 , an explicit set of sparsely intersecting polynomials @xmath389 , and a set of explicit linear families of upper and lower adfs @xmath114 defined by ( [ eq : fadf ] ) .
let @xmath390 be sequences of explicit linear families of continuous adfs where @xmath391 and @xmath392 converge pointwise to @xmath393 and @xmath394 , respectively .
for @xmath395 we define @xmath396 with @xmath230 the lebesgue measure .
note that @xmath397 so is a computable real number as it is the difference of two computable real numbers .
this means that @xmath398 is a computable real number since we are taking finitely many maximums and minimums of computable real numbers .
define @xmath399 we have that @xmath400 is a computable sequence as @xmath398 is a computable real number and @xmath0 is a computably growing basic sequence .
we also have that @xmath401 is a computable sequence since @xmath71 is an explicit sparsely intersecting set of polynomials and by theorem [ thm : tengely ]
. finally , set @xmath402 we will define sequences of integers @xmath403 and @xmath404 inductively .
set @xmath405 clearly , the sequence @xmath406 is computable since the sequences @xmath400 and @xmath401 are computable .
the sequence @xmath407 is computable since @xmath71 is explicit .
thus the sequences @xmath403 and @xmath404 are also computable .
let @xmath408 define @xmath409 by @xmath410 it is easy to show that @xmath411 is a bijection .
put @xmath412 where @xmath413 is a permutation of @xmath414 such that if @xmath415 , then @xmath75
. we can find such a @xmath413 since @xmath71 is sparsely intersecting .
reorder @xmath277 so that it is equal to @xmath416 .
define @xmath417 moreover , define @xmath418 , @xmath419 , and @xmath420 by @xmath421 these functions are defined as @xmath411 is a bijection .
define the sets @xmath422 \cap \mathbb{z}.\end{aligned}\ ] ] note that we can write @xmath423 , where @xmath118 and @xmath119 are sequences of integers .
moreover , @xmath424 we will discuss the computability of these sequences in the proof of main theorem [ thm : main ]
. set @xmath425 we will now work towards proving the second part of main theorem [ thm : main ] .
[ lemma : contadf ] a sequence @xmath426 has a continuous adf @xmath27 if and only if the sequence @xmath427 is u.d .
since @xmath27 is non - decreasing , @xmath428 if and only if @xmath429 . thus @xmath430 , x_n)}{n } = \frac{a_n([0,f(\gamma ) ] , f(x_n))}{n}.\ ] ] but @xmath78 has continuous adf @xmath27 , so @xmath431 , x_n)}{n } = f(\gamma^+)-f(\gamma^- ) = 0.\ ] ] thus @xmath432,x_n)}{n}.\ ] ] similarly we have that @xmath433 , f(x_n))}{n}\ ] ] since @xmath434 ) = 0 $ ] .
therefore @xmath435 , f(x_n))}{n}-f(\gamma).\ ] ] as @xmath27 is continuous , it must map @xmath162 $ ] onto @xmath162 $ ] , so this second limit satisfies the definition for uniform distribution mod @xmath30 .
these limits converge to @xmath436 if and only if the other does , and we are done .
[ lemma : sameadf ] suppose that @xmath0 is a basic sequence that is infinite in limit , @xmath437 is an eventually increasing function , and both @xmath49 and @xmath438 have the same upper and lower adfs .
then @xmath50 has the same upper and lower adfs as @xmath49 .
suppose that @xmath161 is the upper adf of the sequence @xmath50 .
let @xmath439 .
as @xmath27 is eventually increasing , @xmath440 is a basic sequence that is infinite in limit .
it is clear from the definition of @xmath441 that @xmath442 hence @xmath443 we also have that @xmath444 which goes to @xmath436 as @xmath17 goes to @xmath445 . but @xmath446 and @xmath447 have the same upper adf by assumption .
thus for all @xmath448 we have that @xmath449 so @xmath447 has the same upper adf as @xmath450 . the proof for the lower adf is identical
we have approximated each adf @xmath27 by a sequence of continuous adfs @xmath451 as in lemma [ lemma : approxadf ] .
to show @xmath452 has adf @xmath27 we will use lemma [ lemma : contadf ] and lemma [ lemma : sameadf ] as well as the following definition and theorem .
let @xmath453 be a sequence of real numbers .
the * upper discrepancy * with respect to adf @xmath27 of @xmath29 is @xmath454 the * lower discrepancy * of @xmath29 is @xmath455 any theorems or lemmas that hold for the upper discrepancy will also hold for the lower discrepancy with a sign change , so from here on we only state and prove for the upper discrepancy .
many of the properties of the upper and lower discrepancies are the same for the discrepancy , and the proofs are nearly identical .
[ thm : discrep ] the following properties of the upper and lower discrepancies hold for any sequence of real numbers @xmath456 and any continuous adf @xmath27 . 1 . the inequality @xmath457 holds .
2 . if @xmath458 a sequence of real numbers of length @xmath459 , then @xmath460 3 . if @xmath461 , then the upper adf @xmath32 of @xmath29 is bounded above by @xmath27 .
4 . if @xmath55 is a linear family of adfs , then for all @xmath53 , @xmath42 , and @xmath43 , we have that @xmath462 5 . if @xmath32 and @xmath34 are the upper and lower adfs of @xmath29 then @xmath463 .
6 . if @xmath29 is a non - decreasing sequence , then @xmath464 1 .
we have that @xmath465 which implies that @xmath466 for all @xmath448 .
2 . the proof is identical to the proof of theorem 2.6 in chapter 2 of @xcite .
we have that @xmath467 which implies that @xmath468 but this implies that @xmath466 for all @xmath448 .
4 . note that @xmath469 dividing by @xmath53 yields the result .
5 . since @xmath470
, we have that @xmath471 6 .
the proof is identical to the proof of theorem 1.4 in chapter 2 of @xcite .
define @xmath472 [ lemma : morannumber ] for all natural number @xmath17 we have @xmath473 .
the interval @xmath474 is of length @xmath475 .
thus @xmath476 .
but @xmath477 , so @xmath478 . thus @xmath479 .
but @xmath480 , so @xmath481 .
we will use the following basic lemma .
[ lemma : tcorr ] let @xmath482 be a real number and @xmath483 and @xmath484 be two sequences of positive real numbers such that @xmath485 then @xmath486 we now prove the second part of main theorem [ thm : main ] .
the proof that the set @xmath117 has full hausdorff dimension is done similarly .
let @xmath487 , @xmath68 , and @xmath488 , @xmath489 .
the sequence @xmath490 can then be written as @xmath491 where each @xmath492 is a block of digits with length @xmath493 when @xmath494 . when @xmath495 , the length of @xmath492 is less than @xmath493 .
we will look at the discrepancies of the blocks @xmath492 , and apply theorem [ thm : discrep ] to show the claim .
put @xmath496 . by the definition of the upper discrepancy
, we have that @xmath497 unless @xmath498 , in which case the upper discrepancy is bounded by 1 .
we must also consider when the digits of @xmath6 lie along sequences of the form @xmath499 for some other @xmath500 .
for @xmath501 , there are at most @xmath502 polynomials that could intersect @xmath57 , and since @xmath503 at stage @xmath72 , we have that the number of terms that increase the discrepancy is at most @xmath504 . because @xmath505 does not converge to @xmath393 uniformly , we can not measure the discrepancy of the sequence with respect to @xmath393 directly . instead
, we use the fact of pointwise convergence and discrepancy with respect to @xmath505 . defining @xmath506 ,
we can write @xmath507 , and @xmath508 where @xmath509 for this first sum @xmath510 , we find that @xmath511 for the second term @xmath512 we have that @xmath513 for the final term @xmath514 we get @xmath515 thus the sequence @xmath490 has adf bounded above by @xmath393 .
this calculation only shows that the upper adf of the sequence @xmath490 is bounded above by @xmath516 . to show that this is the actual upper adf
, we must find a sequence @xmath517 along which @xmath518 let @xmath519 .
note that for the second term in the previous sum , we have that @xmath520 .
so we need only check that the first term @xmath510 goes to @xmath436 , as @xmath521 and @xmath418 must be @xmath436 for all @xmath17 .
thus by lemma [ lemma : tcorr ] @xmath522 this implies that @xmath490 has upper adf @xmath393 .
note that throughout the proof we only used that @xmath523 , so @xmath524 has the same upper adf as @xmath490 . therefore by lemma [ lemma : sameadf ] , the upper adf of @xmath452 is @xmath393 .
the proof that @xmath490 has lower adf @xmath394 follows similarly . to compute the hausdorff dimension of @xmath116 , we note that @xmath525 is a homogeneous moran set with @xmath526 $ ] , @xmath527 by lemma [ lemma : morannumber ] , and @xmath528 .
so @xmath529 we will now prove that @xmath118 and @xmath119 are computable sequences .
recall that @xmath530 and @xmath531 . to see these sequences
are computable , note that @xmath532 and @xmath533 are computable sequences as @xmath534 and @xmath535 are computable sequences for any polynomial @xmath536 .
thus @xmath537 is a computable sequence . by the explicitness of @xmath538 for each @xmath17 , and uniform computability of the discontinuities of @xmath539
, we have that @xmath540 is a computable sequence .
thus we have that @xmath541 is the ceiling of the product of two computable sequences as @xmath0 is a computably growing basic sequence . hence @xmath542 is a computable sequence .
similarly , it can be shown that @xmath543 is a computable sequence .
[ section : arith ] in the same vein as @xcite , we would like to prove results about the accumulation points of @xmath544 and prove theorem [ thm : strongerwangwenxi ] .
it is difficult to say anything about accumulation points along polynomially indexed subsequences of @xmath544 .
inequivalent polynomials can intersect infinitely often , so the accumulation points of @xmath49 could be difficult to find .
however , if we only sample along arithmetic progressions , a classification of the accumulation points of @xmath48 can be given .
for notational convenience , we define for @xmath545 \to [ 0,1]$ ] , set @xmath546 : f \text { is increasing at } x\}}.\ ] ] after establishing lemmas , we will prove the following theorem .
[ thm : thrm3.2 ] given a basic sequence @xmath0 that is infinite in limit and a linear family of upper and lower adfs @xmath547 and @xmath548 , the set of real numbers @xmath6 such that @xmath131 has upper and lower adfs @xmath549 and @xmath550 and such that @xmath48 has accumulation points equal to @xmath551 has hausdorff dimension 1 .
[ lemma : acc ] if @xmath552 , then @xmath6 is an accumulation point of @xmath48 .
suppose that @xmath552 .
if @xmath553 , then either @xmath554 or @xmath555 for all @xmath317 . the proof for both of these cases
is identical , so suppose that @xmath554 .
then @xmath556 this implies that there are infinitely many @xmath557 for all @xmath317 .
let @xmath558 .
construct a sequence @xmath559 by choosing @xmath560 .
thus @xmath6 is an accumulation point of @xmath48 .
the case where @xmath561 is identical , replacing the @xmath562 with @xmath563 .
if @xmath545\to [ 0,1]$ ] is non - decreasing and @xmath564 , then @xmath565 . since @xmath27 is non - decreasing , if @xmath566 , then @xmath567 . set @xmath568
. then if @xmath569 , we have that @xmath570 and @xmath571 . but this contradicts @xmath6 being the infimum of @xmath572
. thus @xmath573 .
so we have that for all @xmath566 , @xmath573 , and @xmath565 .
let @xmath574 . to see that the set of accumulation points of @xmath48 is exactly @xmath551 ,
note that by lemma [ lemma : acc ] , the set of accumulation points of @xmath48 contains @xmath551 .
for the converse direction , note that by the construction , if @xmath575 is even , then @xmath576.\ ] ] if @xmath575 is odd , then @xmath577.\ ] ] the term @xmath578 goes to @xmath436 . define a sequence @xmath579 with @xmath580 define the sequence @xmath581 similarly . then @xmath582 but @xmath583 . since @xmath584 is a linear family of adfs , we have that this set is a subset @xmath585 , so @xmath586 .
as @xmath585 is closed , all accumulation points of @xmath587 must lie in @xmath585 .
the same statements for @xmath550 are true .
furthermore , we have that by construction @xmath588 is either arbitrarily close to the sequence @xmath589 or @xmath581 , so the set of accumulation points of @xmath48 is contained in the set of accumulation points of @xmath590 .
thus it is a subset of @xmath591 .
let @xmath592 be a topological space and @xmath593 be the measure space of x with the borel @xmath594algebra .
then @xmath595 [ lemma : adfclosed ] for every closed set @xmath183 $ ] , there is an adf @xmath596 so that @xmath597 .
first we construct a measure @xmath598 so that @xmath599 . first consider the borel @xmath600algebra of @xmath186 with the subspace topology .
define @xmath601 extend this measure to the borel @xmath600algebra on @xmath162 $ ] by @xmath602 .
then @xmath599 , @xmath603 .
if @xmath604 , then @xmath605 .
let @xmath606 and @xmath607)$ ] if @xmath608
. then @xmath609 , @xmath610\cap d ) = 1 $ ] , and @xmath596 is non - decreasing since @xmath611)\leq \mu_d([0,y])$ ] if @xmath612 .
if @xmath386 is a set that does not intersect @xmath186 , then @xmath605 and @xmath613 for all @xmath6 , @xmath614 . if @xmath615 with @xmath616 , then @xmath617)\geq y - x>0 $ ]
we will define adfs @xmath618 such that if @xmath619 we will have @xmath620 and if @xmath621 we will have that @xmath622 for some @xmath623 .
applying theorem [ thm : thrm3.2 ] will yield the desired result . for @xmath619 ,
let @xmath624 . for @xmath625
, let @xmath626 define @xmath627 , where @xmath596 is as defined in lemma [ lemma : adfclosed ] . then by theorem [ thm : thrm3.2 ] , the sequence @xmath48 has accumulation points @xmath191 for @xmath623 .
[ section : otherproofs ] the first assertion follows directly from lemma [ lemma : sameadf ] .
for the second , the proof that the set of @xmath0-distribution normal numbers is meagre follows identically to the case of the @xmath1-ary expansion . the fact that the set of @xmath0-distribution normal numbers and the set of numbers @xmath6 where @xmath131 is u.d .
mod @xmath30 have full measure for all @xmath135 follows by a well known result of h. weyl .
thus their difference set has measure zero .
let @xmath628 , @xmath629 , and @xmath630 \hbox { and } x \equiv r\bmod{n}\}.$ ] set @xmath631 $ ]
. for all @xmath17 and @xmath632 , set @xmath633 for @xmath634 , we have that @xmath635 and @xmath636 .
furthermore , for all @xmath53 , we have that @xmath637 . using notation from lemma [ lemma : adfclosed ] ,
define @xmath638 .
thus , we see that @xmath280 is a linear family of adfs . moreover , @xmath639 for all @xmath6 only if @xmath640 and @xmath641 .
thus the set of real numbers @xmath6 where @xmath131 has the adf @xmath101 for all @xmath42 and @xmath43 is a subset of the set of numbers that are @xmath0-distribution normal but ap @xmath0-distribution abnormal and has full hausdorff dimension by main theorem [ thm : main ] .
define the adf @xmath114 as follows .
let @xmath642 be constants such that @xmath643 .
since @xmath166 , we can find an interval @xmath644 such that @xmath645 and @xmath646 .
let @xmath647 be a constant such that @xmath648 .
define @xmath649 then @xmath174 is integrable , so we can apply the lebesgue differentiation theorem to conclude that @xmath650 is differentiable almost everywhere with @xmath651 .
as @xmath174 is non - negative , we have that @xmath114 is non - decreasing .
furthermore , @xmath652 .
since @xmath653 , we have that @xmath654 .
thus @xmath114 is an adf . since @xmath655 almost everywhere , we have for almost every @xmath656 ( respectively @xmath657 ) that @xmath658 ( respectively @xmath659 ) .
thus if @xmath660 is a sequence with adf @xmath114 , we have that @xmath661 thus for any sequence with adf @xmath114 , we have that @xmath662 by main theorem [ thm : main ] , we have that the set of real numbers @xmath6 such that @xmath51 has adf @xmath114 has full hausdorff dimension , so we are done .
let @xmath71 be a set of sparsely intersecting polynomials and let @xmath663 and @xmath664 for every @xmath296 $ ] and @xmath68 .
then @xmath117 has full hausdorff dimension .
moreover , for every @xmath665 , the upper and lower adfs of @xmath452 are not equal .
thus @xmath452 , and in particular @xmath51 , does not have an adf for all @xmath6 on a set of full hausdorff dimension .
the set @xmath117 is of zero measure since for almost every @xmath6 the sequence @xmath51 is u.d .
i. niven and h. s. zuckerman wrote in @xcite : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath666 be a real number with fractional part @xmath667 when written to scale @xmath43 .
let @xmath668 denote the number of occurrences of the digit @xmath1 in the first @xmath17 places .
the number @xmath666 is said to be * simply normal * to scale @xmath43 if @xmath669 for each of the @xmath43 possible values of @xmath1 ; @xmath666 is said to be * normal * to scale @xmath43 if all the numbers @xmath670 are simply normal to all the scales @xmath671 .
these definitions , for @xmath672 , we introduced by mile borel @xcite , who stated ( p. 261 ) that `` la proprit caractristique '' of a normal number is the following : that for any sequence @xmath143 whatsoever of @xmath673 specified digits , we have @xmath674 where @xmath675 stands for the number of occurrences of the sequence @xmath143 in the first @xmath17 decimal places if the number @xmath666 has the property ( [ eq : quote ] ) then any sequence of digits @xmath676 appears with the appropriate frequency , but will the frequencies all be the same for @xmath677 if we count only those occurrences of @xmath143 such that @xmath678 is an @xmath679 digit ?
it is the purpose of this note to show that this is so , and thus to prove the equivalence of property ( [ eq : quote ] ) and the definition of normal number .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ it is not difficult to see how the equivalent definition of normality introduced in theorem [ thm : basebnormalii ] may be confused with the notion discussed in theorem [ thm : basebnormalii ] .
research of the authors is partially supported by the u.s .
nsf grant dms-0943870 .
the authors would like to thank lior fishman and mariusz urbanski for helpful suggestions and reading this manuscript and vitaly bergelson for pointing us to the reference @xcite .
we thank peter mueller and michael zieve for pointing us to the reference @xcite and http://mathoverflow.net/ for providing a forum where we could ask about this .
we are grateful to kostos beros for answering our questions on recursion theory . | we explore in depth the number theoretic and statistical properties of certain sets of numbers arising from their cantor series expansions . as a direct consequence of our main theorem
we deduce numerous new results as well as strengthen known ones . |
solving difficult classical combinatorial problems by adiabatically controlling a fictitious quantum hamiltonian is a fascinating route to quantum computation @xcite , which is alternative to the main - stream approach @xcite .
the quantum annealing ( qa ) strategy @xcite is precisely based on the adiabatic switching off , as a function of time , of strong fictitious quantum fluctuations , suitably introduced by an extra kinetic term in the otherwise classical hamiltonian under consideration . a number of authors have by now applied qa to attack a variety of optimization problems .
some amount of success was obtained in several cases , notably in the folding of off - lattice polymer models @xcite , in the 2d random ising model annealing@xcite , in the lennard - jones clusters optimization@xcite , and in the traveling salesman problem @xcite . despite these successes mostly obtained by path - integral monte carlo ( pimc ) implementations of qa
there is no general theory addressing and even less predicting the performance of a qa algorithm .
different optimization problems are characterized by different `` energy landscapes '' of barriers and valleys about which little is known but which clearly influence the annealing process@xcite .
success of qa crucially depends on the efficiency with which the chosen kinetic energy and associated fictitious quantum fluctuations explores and influences these effective energy landscape .
this is a uncomfortable situation , in view of the fact that it is _ a priori _ not obvious or guaranteed that a qa approach should do any better than , for instance , classical simulated annealing ( ca ) . indeed , for the interesting case of boolean satisfiability problems @xcite more precisely , a prototypical np - complete problem such as 3-sat
a recent study has shown that pimc - qa performs definitely _ worse _ than simple ca @xcite . in a previous paper @xcite , henceforth referred to as i , we compared classical and quantum annealing approaches in their performance to optimize the simplest potential landscape @xmath0 , namely the one - dimensional potential double - well .
classical annealing was implemented by means of a fokker - planck ( fp ) equation with a time - dependent ( decreasing ) temperature @xmath1 .
quantum annealing was performed through the schrdinger s equation , both in real and in imaginary time , describing a particle with a time - dependent ( decreasing ) inverse mass @xmath2 in the double - well potential @xmath0 .
given the textbook simplicity of the problem , it was possible to correlate in detail the features of the landscape ( position and shape of valleys , barrier height ) with the outcome of the annealing process .
one saw for instance that fokker - plank ca is sensitive only to the ratio @xmath3 of the energy splitting @xmath4 between the two wells , or valleys , and the barrier height @xmath5 , whereas on the contrary schrdinger qa is sensitive to the landau - zener tunneling gap , and hence , among other things , to the barrier width .
unfortunately , such a direct approach is applicable , for practical purposes , only to insignificantly small sized optimization problems . to appreciate the difficulty , it is enough to consider that the number of possible configurations , and hence the hilbert space size , of a small 32@xmath632 square lattice ising model is @xmath7 , an astronomically large number which forbids any direct approach based on deterministic state evolution .
hard instances of typical optimization problems involve an even larger number of variables . because of that , alternative strategies , based on monte carlo sampling , are mandatory . that notwithstanding , the various deterministic types of dynamics , fokker - planck , schrdinger , or monte carlo dynamics , classical or quantum , are not at all equivalent
. there is in particular no relationship between the ( physical ) time appearing in the fokker - planck or in the schrdinger equation , and the corresponding classical or quantum monte carlo time - step .
furthermore , _ many _ different types of monte carlo ( mc ) dynamics are possible .
while sampling the same equilibrium probability distribution and hence providing the same equilibrium averages @xcite , they will generally perform very differently in an out - of - equilibrium situation .
in order to appreciate that , given the large variability of the possible mc outcomes , it is once again wise to concentrate attention on landscapes which are well under control . in this paper
we implement and investigate different monte carlo annealing strategies , both classical and quantum , to the very same optimization problem the one - dimensional double - well which was studied in i by deterministic ( fokker - planck and schrdinger ) annealing approaches .
we will first show ( sec .
[ ca_intro : sec ] that the choice of the proposed move in a metropolis mc scheme strongly influences the annealing performance .
such a strong influence of the move is seen to be correlated with the instantaneous spectrum of the markov transition matrix associated to the mc scheme , as shown in detail in sec .
[ spectral : subsec ] .
next , we will turn to quantum mc , specifically to path - integral monte carlo ( pimc ) , and show how the required tunneling dynamics is highly non - trivial .
the outcome of an allegedly state - of - the - art pimc annealing of a simple double - well can even be very unsatisfactory , due to sampling difficulties of instanton events ( see secs .
[ pimc : sec ] , [ results1:subsec ] , [ results2:subsec ] ) .
finally , we will analyze the choice of kinetic energy in the qa hamiltonian , by testing the effectiveness of an alternative _ relativistic _ kind of dispersion , which proves much more effective than the usual non - relativistic choice ( sec .
[ results3:subsec ] ) .
section [ discussion : sec ] will finally contain a summary of the main results some discussion .
technical details on the spectral analysis of the classical markov process , and on the generalization of the bisection algorithm of interest for the pimc study of sec .
[ results3:subsec ] , are relegated to the two appendices .
let us start with the double - well one - dimensional potential on which we will test the different classical and quantum monte carlo ( mc ) annealing schemes presented in this paper .
we assume @xmath0 to describe two generally asymmetric wells of the form : @xmath8 where @xmath9 ( both positive ) , and the linear term @xmath10 , with @xmath11 , splits the degeneracy of the two minima .
( the discontinuity in the second derivative at the origin is of no consequence in our discussion . )
the optimal value of the potential is obviously @xmath12 . to linear order in the small parameter @xmath13 ,
the two minima are located at @xmath14 , and the splitting between the two minima is @xmath15 , while the second derivative of the potential at the two minima , to lowest order in @xmath13 , is given by @xmath16 .
a `` symmetric '' double - well potential is recovered for @xmath17 .
starting with some initial state or distribution , the key quantity under inspection will be the average residual energy @xmath18 after annealing for a certain time @xmath19 , where @xmath20 is the average final potential energy . in the classical mc case
, annealing is performed by reducing the temperature @xmath21 . as shown in i ( ref . ) , the results of a fokker - planck annealing are roughly independent of the asymmetry of the potential , and only controlled by the ratio @xmath3 of the splitting @xmath4 between the two minima and the barrier height @xmath22 . in fig .
[ fp_vs_mc_sym_fdt : fig ] we present the results of a standard metropolis mc classical annealings for a symmetric ( @xmath23 , @xmath24 ) double - well potential , where a linear term @xmath10 with @xmath25 provides a splitting between the two minima of order @xmath26 .
entirely similar results ( not shown ) are obtained for asymmetric double - wells .
the solid triangles represent the results of exact integration of fokker - planck s ( fp ) equation , via a fourth - order runge - kutta method , with a linear annealing schedule for the temperature @xmath21 , as reported in i. with @xmath23 , @xmath24 , and a linear term @xmath25 .
the initial condition was @xmath27 .
the solid line through the points is a fit with a power - law of leading exponent @xmath28 times logarithmic corrections , see i. clearly , the fokker - planck and the monte carlo annealing `` times '' are not related ( see text).,width=302 ] the circles , squares , and diamonds represent the results of three different classical metropolis mc annealings clearly showing very different behaviors , differing uniquely in the choice of the proposed move . to clarify this point , we recall that the markov process associated with a metropolis mc is fully specified by a transition probability @xmath29 , to move from @xmath30 to @xmath31 in the time step @xmath32 .
w is the product of two factors : @xmath33 where @xmath34 is the probability of _ proposing _ a move from @xmath30 to @xmath31 , while @xmath35/k_bt } \right\ } \;,\ ] ] is the probability of _ accepting _ the proposed move .
the various results shown in fig . [ fp_vs_mc_sym_fdt : fig ] differ in the choice of @xmath34 , i.e. : + 1 ) ( box , circles ) the new proposed position @xmath31 is uniformly distributed inside a box of length @xmath36 centered around @xmath30 , i.e. , @xmath37 2 ) ( gaussian , squares ) the new position @xmath31 is distributed according to a gaussian of width @xmath38 centered around @xmath30 , i.e. , @xmath39 3 ) ( lorentzian , diamonds ) the new position @xmath31 is distributed according to a lorentzian of width @xmath38 centered around @xmath30 , i.e. , @xmath40 in all three cases , the value of @xmath38 , which controls the move range , was decreased with temperature according to @xmath41 , with @xmath42 , for the box and the gaussian case , and @xmath43 for the lorentzian one .
this choice was made empirically by monitoring the average acceptance and the root mean square displacement , @xmath44 , of the mc process . as evident from fig .
[ fp_vs_mc_sym_fdt : fig ] , the results of the various choices of proposal moves differ in a substantial way .
while the box choice leads to a residual energy very close to the deterministic fokker - planck result , the gaussian choice , and even more so the lorentzian one , lead to residual energies which decrease faster with the annealing time @xmath19 .
the reason for the non uniqueness of the result of a classical mc annealing is not difficult to grasp .
the metropolis algorithm guarantees that the expectation values of a system at equilibrium are correctly reproduced by a mc time average , after a suitable equilibration time @xmath45 .
the equilibration time @xmath45 depends on the quantity one is interested , and , more crucially for our scope , on the type of proposed moves , i.e. , on the @xmath34 . for equilibrium mc simulations ,
this issue is of little concern : as long as one checks that the stochastic process has reached a stationary condition , the results are guaranteed to be correct . in an annealing simulations , however , ( or , in that respect , in any out - of - equilibrium evolution )
, the choice of the @xmath34 , determining the instantaneous equilibration time of the process , strongly influences the overall performance of the algorithm . evidently , both the gaussian and the lorentzian trial moves allow the attainment of residual energies @xmath46 much below that of the box move ( and of the fp dynamics ) .
the lorentzian case is particularly impressive .
if we fit the asymptotic part of the residual energy data by @xmath47 , the annealing exponent @xmath48 turns out to be @xmath49 for the gaussian case , and @xmath50 for the lorentzian one , i.e. definitely much larger than what the huse and fisher theory @xcite predicts for the fokker - planck case , namely @xmath51 .
the long lorentzian tails clearly imply an abundance of moves operating _ long jumps _ , which are beneficial to the annealing even at small temperatures @xmath52 .
indeed have reported a genuine improvement of the sampling by making use of the lorentzian distribution , instead of the gaussian , for the thompson problem ( finding the equilibrium distribution of @xmath53 charged particles on a sphere ) @xcite , and in optimizing the structure of a cluster made of @xmath53 ni atoms @xcite . in order to analyze these issues in more quantitative detail
, we carried out a spectral analysis of the transition matrices @xmath29 corresponding to the different proposal move schemes . in the following section we briefly recall the crucial steps behind this approach .
every discrete - time markov process can be defined by means of an equation of the form : @xmath54 where , for simplicity of notation , we have assumed that the possible states of the system , labeled @xmath55 and @xmath56 , are also discrete ( we will indeed reduce our continuous problem to a discrete one in the following numerical calculations , see below ) .
@xmath57 is here the probability of finding the system in state @xmath55 at time @xmath58 , and @xmath59 is the so - called _ transition matrix _ ,
i.e. , the transition probability for going from state @xmath56 to state @xmath55 during the time step @xmath32 .
the transition matrix @xmath59 ( assumed to be time - independent ) is a stochastic matrix @xcite , in the sense that @xmath60 and @xmath61 . in our particular case ,
the stochastic process behind the metropolis sampling is given by eq .
[ metropolis_t : eqn ] , so that @xmath59 also fulfills the detailed balance condition @xcite .
the idea underlying the spectral analysis of @xmath59 is very simple . for a fixed temperature @xmath52 , the markov process admits a spectral analysis in much the same spirit in which one analyzes a time - dependent schrdinger equation via the spectrum of the associated hamiltonian .
indeed , in continuous time
i.e. , for @xmath62 the result is quite simple to state .
a similar analysis can be carried out for the original discrete - time markov chain , see ref . .
as explained there , from @xmath59 one can define a closely related matrix @xmath63 appearing in the short - time expansion , @xmath64 , of the exponential map : @xmath65 ; in turn , @xmath66 possesses a set of right eigenvectors @xmath67 with @xmath68 ( the equilibrium distribution ) , and the corresponding eigenvalues @xmath69 are such that @xmath70 , @xmath71 @xmath72 . in terms of the spectrum of @xmath73 , one can formally write the general solution of the time - continuum ( @xmath62 ) limit of eq .
( [ markovprocess ] ) as @xcite : @xmath74 where @xmath75 are real coefficients depending on the initial ( @xmath76 ) condition .
[ spectral_decomposition : eqn ] shows that the approach to the equilibrium distribution is governed by a _ relaxation time _
@xmath77 , @xmath78 being the smallest non - zero eigenvalue of @xmath66 ( recall that @xmath70 )
. therefore , the larger the _ spectral gap _
@xmath79 , the faster the system reaches equilibrium .
however , since in the metropolis algorithm the time step is actually discrete ( see appendix [ model_spectrum : sec ] ) , it is better to consider the set of the @xmath80 eigenvalues , given by @xmath81 , with @xmath82 . therefore , in the rest of the paper we shall refer to the spectral gap of the original transition matrix @xmath80 , which is @xmath83 , instead of the corresponding time - continuum limit introduced above . as we shall show in a while , this choice will not change the conclusion of our spectral analysis . how can we use the spectral gap concept in an annealing context ?
it is natural to think that , for a given schedule @xmath1 , it would be good to _ maximize _ the spectral gap for each given temperature @xmath1 .
this is indeed what we will explore in the following . generally speaking , finding explicitly the spectral gap @xmath84 for a given @xmath73 is an impossible task ; in the present one - dimensional double - well context , however , it turns out that the spectrum of @xmath85 is quite simple to obtain by , for instance , a straightforward discretization of the real axis .
we present now the results obtained by applying this spectral analysis to the proposal moves we defined in sec .
[ ca_intro : sec ] .
we recall that they are : the box move , see eq .
[ p_transition_matrix_box : eqn ] , the gaussian move , see eq .
[ p_transition_matrix_gau : eqn ] , and the lorentzian one , see eq .
[ p_transition_matrix_lor : eqn ] .
the main quantity under inspection is the _ spectral gap _
@xmath84 of the transition matrix @xmath85 which is a function of both the temperature @xmath52 and the proposal move range @xmath38 , @xmath86 .
we obtain this gap ( as well as the higher eigenvalues ) by diagonalizing an appropriate symmetrized ( see appendix [ model_spectrum : sec ] ) and discretized version of the transition operator @xmath85 .
@xcite , for the symmetric potential case .
we plot the spectral gap @xmath84 versus the proposal move range @xmath38 .
the temperature is @xmath87 ( top ) or @xmath88 ( bottom ) , in the same units in which the barrier parameter is @xmath24 .
the solid and dashed black lines represent the fit provided by eq .
[ theory_gap_gau : eqn ] ( gaussian case ) and eq .
[ theory_gap_lor : eqn ] ( lorentzian case ) .
the functional form of those fits is recalled in the legend.,width=302 ] in fig .
[ all_vs_sigma_tot : fig ] we plot the spectral gap @xmath84 versus @xmath38 at two fixed temperatures , @xmath87 ( top ) and @xmath88 ( bottom ) . as a general remark ,
we note how the behavior of @xmath89 is remarkably different for different trial moves .
interestingly , for every proposal move and every @xmath52 , there is a value of @xmath38 which maximizes the gap @xmath84 . recalling that the relaxation time is proportional to @xmath90 we anticipate that , at every temperature @xmath52
, there is a @xmath91 which provides the fastest relaxation .
clearly this @xmath91 is _ a priori _ a better choice than taking just @xmath92 , as previously implemented in sec .
[ ca_intro : sec ] . at low @xmath52
however the dependence of @xmath84 on @xmath38 is no longer smooth . in particular , for @xmath88 , the box trial move shows a sharp transition from vanishingly small gap values to finite ones , while both gaussian and lorentzian moves show a cusp maximum . at low temperatures ,
the maximum gap @xmath93 of the lorentzian move is larger than @xmath94 , which is in turn larger than @xmath95 ; correspondingly , for small @xmath38 , we have @xmath96 .
this is , essentially , the reason for the faster relaxation dynamics of the lorentzian move during annealing .
notice also that the maximum value @xmath97 of @xmath84 , decreases with temperature in all cases , as one can see by comparing the two panels in fig .
[ all_vs_sigma_tot : fig ] .
moreover the value of @xmath38 that maximizes the lorentzian and gaussian gap decreases with temperature , while it seems to converge to a finite value ( around @xmath98 ) for the box case .
, for the symmetric potential , with box proposal move .
we plot the spectral gap @xmath84 versus the inverse temperature @xmath99 for several fixed values of the proposal move range @xmath38 .
top left inset : the value of the effective barrier , @xmath100 , seen by the system when a box proposal move with range @xmath38 is employed .
the value @xmath101 which corresponds to a vanishing effective barrier is indicated by a vertical arrow .
top right inset : the two first gaps , @xmath102 and @xmath103 , for a box proposal move versus the inverse temperature , @xmath99 , at fixed move range @xmath104,width=302 ] .
[ box_vs_t : fig ] the box case is particularly intriguing , and deserves a closer inspection . in fig . [ box_vs_t : fig ] we plot @xmath84 versus inverse temperature @xmath99 for the case of the box move , and for several values of @xmath38 . for a whole range of @xmath38 , @xmath105
, we see a low - temperature behavior typical of an activated arrhenius process @xmath106 , where @xmath107 is an _ effective barrier _ , clearly @xmath38-dependent , which the system experiences . above a certain @xmath101 ,
the behavior of @xmath84 changes drastically , from arrhenius to what appears , at first sight , just a constant .
a closer inspection on an extended temperature range , see fig .
[ box_vs_t : fig ] ( top right inset ) , shows that at very small temperatures an ( avoided ) crossing between eigenvalues of @xmath80 has taken place , and @xmath84 starts to decrease again toward zero as @xmath108 .
@xcite the effective barrier @xmath107 extracted from an arrhenius fit , @xmath109 , goes to zero for @xmath110 .
there is a simple geometrical explanation for this critical behavior .
first of all , we note that the value of @xmath101 is close to the distance between the two minima .
this suggests that the transition is related to the availability of a _ direct jump _ from the bottom of the metastable well to the other one .
if we call @xmath111 , respectively , the minimum in the right and left well , there is a non - trivial solution of the equation @xmath112 with @xmath113 lying between the two minima , @xmath114 ( see fig .
[ fig_double_well_barrier : fig ] ) : it turns out that @xmath115 . , while in the lower one we illustrate the general case @xmath116 . here
@xmath117 is the potential at the bottom of the metastable minimum .
the meaning of the other symbols is explained in the text .
, width=302 ] indeed , for @xmath118 there is a possible proposed move that brings the system from @xmath119 ( the metastable minimum ) to the point @xmath120 on the other side of the barrier : such a `` non - local '' move pays no energy ( @xmath121 ) , and is therefore certainly accepted by the metropolis algorithm . in some sense , for @xmath122 the barrier is not seen ( or , at least , there are allowed and accepted moves that do not see it ) , and @xmath100 is zero .
consider now the case @xmath123 .
for every value @xmath124 , with @xmath125 , there are two equipotential solutions @xmath126 and @xmath127 , such that @xmath128 , lying between the two minima , but on opposite sides of the barrier , i.e. , with @xmath129 .
denote now by @xmath130 the distance between such two equipotential points ( @xmath131 is a monotonically decreasing function of @xmath132 ) . if the box width is @xmath38 , we can always find @xmath132 such that @xmath133 , i.e. , such that there is a proposed move connecting @xmath134 to @xmath135 which , once again , bypasses the top of the barrier .
the effective barrier seen by the system close to the metastable minimum , for a given value of @xmath38 , is therefore just ( see fig .
[ fig_double_well_barrier : fig ] ) @xmath136 i.e. , in essence , a piece of length @xmath38 is cut from the top of the barrier , and is effectively not seen by the system . in other words ,
the effective barrier @xmath100 is just the potential energy drop @xmath132 which the system must overcome before a long jump @xmath137 is made available at no energy cost ( @xmath138 ) .
the value of @xmath107 obtained through such a simple geometric construction is shown by a solid line in fig .
[ box_vs_t : fig](top left inset ) : the agreement with the numerical data extracted from the arrhenius fit is remarkably good , with small deviations for very small values of @xmath38 , which are likely due to the effect of the finite grid employed in diagonalizing the transition operator @xmath80 ( see sec .
[ spectral : subsec ] ) , and to finite @xmath52 effects .
@xcite we would like stress that this geometrical picture is strictly true only at zero temperature , and does not apply to the gaussian or lorentzian cases , whose tails provide a small non - vanishing chance of making a `` long jump '' for any value of @xmath38 .
the treatment of the gaussian and lorentzian cases needs , therefore , a more specific and technical discussion , which we sketch in appendix [ model_spectrum : sec ] .
summarizing , the box trial move shows a _ sharp _ ( first - order - like ) transition at a value of @xmath139 , where the effective arrhenius barrier @xmath107 vanishes , and the gap starts to decrease as a power - law , @xmath108 , for very small values of @xmath52 ( see fig .
[ box_vs_t : fig ] ) .
one might suspect that the cusps shown in fig .
[ all_vs_sigma_tot : fig ] for the gaussian and lorentzian cases signal , similarly to the box case , some kind of transition . a closer inspection , however , shows that this is not so : it is just a level crossing phenomenon . versus the inverse temperature @xmath99 for several fixed values of the proposal move range @xmath38.,width=302 ] in fig .
[ gau_plus_lor_vs_t : fig ] we plot @xmath84 versus the inverse temperature @xmath99 for several values of @xmath38 , for the gaussian ( top panel ) and lorentzian ( bottom panel ) proposal moves . after an initial arrhenius - like behavior ,
particularly visible for the gaussian small @xmath38 cases , the system always , and smoothly , changes to a low-@xmath52 behavior which is entirely similar to the large-@xmath38 box case : an apparent saturation of @xmath84 followed by an avoided crossing between the first two exited states ( not shown ) and a final @xmath108 .
@xcite no real transition exist in the gaussian and lorentzian case : the cusp in fig .
[ all_vs_sigma_tot : fig ] moves down toward smaller and smaller values of @xmath38 for decreasing @xmath52 . in the previous section we argued about the possibility of an optimal choice for the proposal move range @xmath91 .
this choice should guarantee the fastest _ instantaneous _ relaxation toward the instantaneous equilibrium distribution @xmath140 at any given temperature .
the classical annealing performance is expected to be greatly improved by such a choice of @xmath91 . in practice , for each choice of proposal move ( box , gaussian , lorentzian ) , we numerically determined the value of @xmath38 that maximizes the spectral gap @xmath84 for each fixed temperature @xmath52 .
we will refer here to such an optimal @xmath38 as @xmath141 . for the box , gaussian and lorentzian proposal move .
we also show , for comparison , the schedule @xmath142 employed in sec .
[ ca_intro : sec ] .
, width=302 ] in fig .
[ fig_optimal_sigma_vs_t : fig ] we plot such an optimal choice for all the types of proposal moves introduced in sec .
[ ca_intro : sec ] .
for comparison , we also show the @xmath91 we employed in sec .
[ ca_intro : sec ] , which is definitely smaller than @xmath143 , for given value of @xmath52 .
having done this , we have then performed classical mc annealing runs , similarly to those reported in sec .
[ ca_intro : sec ] , with the usual linear annealing schedule for the temperature @xmath144 , where @xmath19 is the annealing time . is the annealing time and @xmath145 the residual energy ( see text ) .
we report the results of the exact integration ( fokker - planck ) , together with the actual mc data for several proposal moves ( box , gaussian , lorentzian ) .
the mc simulations are performed with an optimal choice for @xmath91 obtained from the maximum instantaneous gap ( see text).,width=302 ] in fig .
[ fp_vs_mc_sym_opt : fig ] we show the mc annealing results .
we stress again that the difference with respect to the runs illustrated in fig .
[ fp_vs_mc_sym_fdt : fig ] , where we took @xmath146 , is that , here , @xmath141 is the optimal choice of @xmath38 obtained from the maximum instantaneous spectral gap of the transition matrix .
we notice that the box and gaussian results are now very different form the previous ones ( see fig .
[ fp_vs_mc_sym_fdt : fig ] ) .
moreover , interestingly , box and gaussian data fall almost on the same curve , which is in turn very close to the lorentzian case , the latter showing once again the best annealing results
in i ( ref . ) we studied , by direct numerical integration , the real- and imaginary - time schrdinger dynamics provided by the _ time - dependent _ hamiltonian @xmath147 where @xmath148 = \gamma_0\,\left ( 1-\frac{t}{\tau } \right)$ ] is the inverse mass parameter appearing in the usual non - relativistic kinetic energy , providing the strength of the quantum fluctuations .
this is a literal implementation of the quantum annealing ( qa ) strategy .
however as pointed out in the introduction , this schrdinger dynamics is suitable only for toy problems with a very limited hilbert space , and quite inapplicable for actual optimization . in order to be a viable strategy for realistic optimization
, qa must resort to _ stochastic _ , i.e. , quantum monte carlo ( qmc ) , approaches , mostly appropriate for an _ imaginary - time _ framework ( we remind the reader that , as shown in i , working in imaginary - time is actually beneficial for qa ) .
there are several possible qmc techniques on which a qa strategy can be build . by far ,
the simplest of these qmc techniques is the _ path - integral monte carlo _ ( pimc ) method , which has already been used with some success in the qa context @xcite .
the method does not addresses the imaginary - time schrdinger time - dependent dynamics , but simulates an _ equilibrium quantum system _ , held at a small finite temperature @xmath52 , where the relevant quantum parameter @xmath149 is externally switched off in the course of a qmc simulation : @xmath58 is not treated , therefore , as a proper physical time , and is only replaced by a _
monte carlo time_. @xcite in the present section we will explore the potential of a pimc - based qa strategy on the same simple one - dimensional double - well potential .
the reason for investing so much effort on this simple problem , is that the simplicity of the problem landscape will allow us to have a perfect control of all the ingredients of the method .
indeed , we will learn a lot about pimc - qa , particularly about the limitations of the method , from investigating this toy problem .
the equilibrium properties of our quantum system , for any fixed value of @xmath150 , are all encoded in the quantum partition function @xmath151 = \int dx \ ; \langle x| e^{-\beta(\hat{t}+\hat{v } ) } |x \rangle\ ] ] where @xmath152 from which all the thermodynamics follows . here , as usual , @xmath153 .
we will drop the boltzmann constant @xmath154 from now on . as we see from eq .
[ partitionfunction : eqn ] , @xmath155 involves an integral over a positive distribution , @xmath156 , which involves , however , the very difficult task of calculating the diagonal matrix element of the _ exponential _ of the hamiltonian @xmath157 .
the standard approach is to rewrite eq .
[ partitionfunction : eqn ] as a feynman _ path - integral _ @xcite .
the main mathematical tool employed in such an approach is the so - called _
trotter theorem _ , which reads : @xmath158 where @xmath159 is the number of _
trotter s slices ( or replicas ) _ , in which the imaginary time interval @xmath160 $ ] has been partitioned , each slice being of length @xmath161 . by means of eq .
[ trotter_theorem : eqn ] , and inserting @xmath162 identities in the form of @xmath163 , one can rewrite the partition function ( see , for instance , ref . for this simple derivation ) as : @xmath164 } + o\left(\delta\,t^2\,\beta\right ) \;,\ ] ] where @xmath165 $ ] is the so - called ( euclidean ) _ primitive action _ : @xmath166 = \delta\,t\,\sum_{i=0}^{p-1}\ , \left\ { \frac{m}{2}\,\left ( \frac{x_{i+1}-x_i}{\delta\,t } \right)^2 + v(x_i ) \right\ } \;,\ ] ] with periodic boundary conditions @xmath167 , as a consequence of the trace present in eq .
[ partitionfunction : eqn ] . )
the similarity of eq .
[ primitive_approx_text : eqn ] with the classical partition function of a _ closed polymer _ with @xmath159 beads is evident : the polymer beads @xmath168 can be seen as the imaginary - time snapshots @xmath169 , @xmath170 , of a fluctuating closed path @xmath171 in the enlarged configuration space @xmath172 , where @xmath58 is the imaginary time .
two neighboring beads @xmath168 and @xmath173 interact with harmonic interactions of spring constant @xmath174 , originating from the propagator of the quantum kinetic term , @xmath175 the strength of the harmonic interactions in imaginary - time controls the amount of quantum fluctuations : a large mass @xmath176 ( classical regime ) results in a strong @xmath177 and hence in a very `` rigid '' polymer , while a small mass @xmath176 ( quantum regime ) results in a very soft and fluctuating polymer .
all beads are also subject to the classical potential @xmath0 . once we have reduced our problem to the path - integral form in eq .
[ primitive_approx_text : eqn ] , the standard techniques of classical metropolis mc can be used , and the resulting algorithm is what is called a _ path - integral monte carlo _ ( pimc ) .
the most obvious mc moves to be used are just _ single bead _
moves , exactly as in a classical mc .
these are the bare bones of a pimc approach . for
the problem we are dealing with , a particle in a potential ( or , more generally , for systems of quantum particles on the continuum ) , one can improve on the method just sketched in two possible directions : _
i ) _ by improving the quality of the approximation in eq .
[ primitive_approx_text : eqn ] , so as to get a smaller trotter discretization error @xcite for a given number of trotter slices @xmath159 used ; @xcite _ ii ) _ by introducing mc moves which are more sophisticated than just moving a single bead at a time . regarding _
i ) _ , we will adopt a _
fourth order _ approximation to the action @xcite , which improves the trotter truncation error of the partition function from @xmath178 to @xmath179 .
more precisely , we used the so - called _ takahashi - imada _
approximation @xcite , which is especially suitable for continuous systems .
the partition function is still given by an expression of the form eq .
[ primitive_approx_text : eqn ] with the primitive action @xmath180 replaced by the following takahashi - imada action : @xmath181 = \delta\,t\ , \sum_{i=0}^{p-1}\ , \left\ { \frac{m}{2}\ , \left ( \frac{x_{i+1}-x_i}{\delta\,t } \right)^2 + v_{\rm eff}(x_i ) \right\ } \;,\ ] ] where the only difference with respect to @xmath180 is in the potential energy , which now reads : @xmath182 as for the mc move choice , _ ii ) _ , we adopted smarter moves , known in the context of classical polymer simulations , which reconstruct entire pieces of the polymer , instead of a single bead at a time , through the _ bisection algorithm _ @xcite .
this choice guarantees a fast relaxation to the instantaneous equilibrium distribution .
moreover , we also used global classical moves , in the form of rigid motions of the center of mass of the polymer @xcite , so as to keep a good sampling in the final part of the annealing , where the mass @xmath176 is large ( @xmath183 is small ) and quantum moves are suppressed . in the next section , we will present the annealing results obtained with such an allegedly _ state - of - the - art _ pimc .
quantum annealing is performed by decreasing the inverse - mass parameter , @xmath184 $ ] , appearing in eq .
[ quadratic_hamiltonian : eqn ] , linearly to zero in a time @xmath19 , @xmath185 ( it is understood here that both @xmath58 and the total annealing time @xmath19 are mc times , i.e. , measured in units of mc steps , each mc step consisting of one bisection move plus one global move ) .
the initial condition is set to @xmath186 ( as in i ) . for every value of the annealing time @xmath19 we calculated the relevant averages by repeating several times the same annealing experiment , starting from different
randomly distributed initial conditions .
@xcite in fig .
[ se_vs_mc_nolor_noins : fig ] we plot the final pimc - qa residual energy obtained for both potential choices , asymmetric ( top ) and symmetric ( bottom ) ( with the same parameters used in i and in sec .
[ ca_intro : sec ] ) , for a fixed temperature @xmath187 .
the statistical errors are evaluated by @xmath188 repetitions of each annealing run for @xmath189 , and @xmath190 repetitions for @xmath191 .
we employed @xmath192 with an @xmath193 bisection level ( i.e. , moving polymer pieces containing @xmath194 beads @xcite , for @xmath195 , and @xmath196 with @xmath197 ( moving @xmath198 beads ) for @xmath191 .
we stress that the two series of data match perfectly , but obviously the use of a smaller @xmath159 allows us to achieve larger annealing times @xmath19 , which would be otherwise too computationally heavy .
( an analysis of the convergence behavior of the algorithm as a function of both @xmath183 and @xmath159 can be found in ref . . ) , for @xmath187 . as a reference , the results obtained in ref . by exact integration of the imaginary - time schrdinger equation
are reported .
( bottom ) same as above , for a symmetric potential.,width=302 ] in the asymmetric potential case ( top panel of fig .
[ se_vs_mc_nolor_noins : fig ] ) which , we recall , presents a level crossing and a landau - zener ( lz ) transition ( see i ) , the residual energy @xmath46 as a function of the annealing time @xmath19 gets stuck around @xmath199 . since
this energy is comparable to the energy of the metastable state , @xmath200 , and definitely larger than the thermal limit @xmath201 , we see that the algorithm failed to follow adiabatically the ground state . for comparison , we also reported in fig .
[ se_vs_mc_nolor_noins : fig ] the residual energy data obtained by the exact imaginary - time schrdinger annealing ( reported in i ) , even if the time - scales of the two algorithms are definitely different and unrelated . considering , on the other hand , the symmetric potential case , bottom panel of fig .
[ se_vs_mc_nolor_noins : fig ] , we notice that the algorithm succeeds in reaching the thermal limit @xmath202 in a reasonable amount of mc steps . from this first comparison between pimc - qa and exact schrdinger
qa , we appreciate that the lz transition leads to a _ severe slowing down _ of the monte carlo algorithm , the actual tunneling event being essentially missed by the pimc algorithm .
this difficulty is also present in _ static _
simulations performed in the neighborhood of the lz crossing ( occurring at @xmath203 ) , where we find ( results not shown ) a dangerous loss of adiabaticity which calls for an improved sampling of the action : we will show how this improved sampling is achieved by the introduction of _ instanton moves_. in the previous section we tested an allegedly state - of - the - art pimc - qa algorithm for the very simple problem of a particle in a double - well potential , with disappointing results .
the problem is a _ sampling crisis _ : our action is accurate , but its sampling misses the tunneling events , which is catastrophic when a landau - zener crossing occurs ( asymmetric potential case ) . the well - known cure for this kind of sampling problem , for the case of a _ perfectly symmetric _
double - well , @xmath204 , is the introduction of the so - called _ instanton move _
( see ref . ) . in a nutshell ,
an instanton is defined as a solution of the classical equation of motion in the inverted potential , which goes from one minimum to the other .
moreover , the time - reversed path ( anti - instanton ) is also a solution .
instanton solutions are given by : @xmath205 where the @xmath206 signs denote instanton and anti - instanton , respectively , @xmath207 is the instanton frequency , and @xmath208 a configuration parameter ( the instanton center ) .
strictly speaking , the whole classical trajectory takes an infinite ( real ) time , while the barrier overcoming is a very fast process ( whence the name instanton ) . in our implementation
, we made use of the imaginary - time version of an instanton / anti - instanton pair as a proposal mc move @xcite .
in practice , the instanton move proposes to a subset of the trotter s slices an excursion from one minimum to the other .
the move will be accepted according to the usual metropolis criterion , with an energy competition between the possible gain in potential energy compensated by the increase in kinetic energy due to the spring stretchings ( at the positions of the instanton and of the anti - instanton ) .
obviously , the instanton move will be not effective in the classical limit ( small @xmath183 , final part of the annealing ) , since it describes an inherently quantum effect : the tunneling process between the two wells .
in particular , the instanton frequency , @xmath209 ( defined in eq .
[ instanton_def : eqn ] ) , decreases with @xmath183 , and when @xmath210 , the tunneling time will exceed the thermal cut - off . as a consequence , the instanton / anti - instanton pair will reduce to a useless flat configuration over the interval @xmath211 $ ] . before showing the results ,
we need to stress an important point : the instanton move is available only for potentials which are small deformations of a perfectly symmetric double - well potential as our symmetric and asymmetric cases are , see sec .
[ ca_intro : sec ] and is not the general key for solving sampling problems ( ergodicity breaking ) for a generic potential , whose landscape is generally poorly known .
in essence , we are playing here , for demonstration purposes , an unfair game , using a _ detailed information _ on the potential landscape in order to correctly implement the tunneling dynamics in our pimc .
[ se_vs_mc_nolor : fig ] shows the pimc - qa residual energy results when the instanton move is introduced , for the case of asymmetric ( top ) and symmetric potential ( bottom ) .
as in the previous section , we employed @xmath192 trotter s slices with @xmath193 bisection level , for @xmath195 , and @xmath196 with @xmath197 bisection level for @xmath191 .
statistical errors are evaluated with @xmath188 repetitions of every annealing run . , with the instanton move allowed . as a reference
the results obtained without instanton move are still reported .
the data fit for the asymmetric potential case is discussed in the text.,width=302 ] we note how the introduction of the instanton move causes a visible improvement of the residual energy slope for the asymmetric potential case : the asymptotic power - law behavior is now quite evident , @xmath212 , although with an exponent which is only @xmath213 ( smaller than in the schrdinger integration case , where @xmath214 , see i ) . for the symmetric potential case , the instanton move leads to a faster convergence to the thermal limit , but does not really give an overwhelming qualitative change .
in particular we note that the partial acceptance ratio of the instanton moves is very low around @xmath215 , despite the fact that this proposal move was `` tailored '' on the potential under investigation .
nevertheless , the instanton move quantitatively changes the pimc - qa performance .
this is perhaps catch a generic feature of systems with barriers , namely the crucial importance of rare events tunneling in this case .
unfortunately , as said , the instanton `` recipe '' , natural as it is for a double - well potential , can not be generalized to generic landscapes , including more complicated potentials , not to mention actual combinatorial optimization problem .
nevertheless , the results obtained are instructive in many respects : they show clearly an important limitation of the pimc approach , while demonstrating , once more , that a smart choice of the proposal move even in the case of pimc - qa leads to immediate improvement of the annealing performance .
it is by now a sort of _ leitmotiv _ of the paper , that the choice of the mc move strongly influences the dynamics , and hence the annealing behavior .
we recall , in particular , see sec .
[ ca_intro : sec ] , that in the classical annealing case the winning choice was given by a lorentzian distributed proposal move , which sometimes provides very long displacements . in the present pimc context , however , the non - relativistic form of the kinetic energy part of the hamiltonian eq .
[ quadratic_hamiltonian : eqn ] _ forces _ , in a sense , the choice of gaussian distributed moves , because the free propagator of the non - relativistic kinetic energy is a gaussian , see eq .
[ hkin_primitive : eqn ] .
the whole bisection algorithm makes strong use @xcite of the gaussian nature of the free propagator .
it is natural to ask what would be the qa behavior if , instead of the standard non - relativistic kinetic energy used so far , we employ for example a _ relativistic _
hamiltonian of the form @xmath216 one immediate consequence of this choice is that the free gaussian propagator appearing in eq .
[ hkin_primitive : eqn ] is now replaced by a lorentzian : @xmath217 with a certain effort , we have succeeded in generalizing the bisection algorithm to implement in a very effective way the dynamics provided by this lorentzian propagator ( see appendix [ lorentzian_move : sec ] and ref . for more details ) .
we present here the results of a pimc - qa approach which implements the relativistic kinetic energy , through a bisection algorithm with such a `` lorentzian move '' . in fig .
[ se_vs_mc : fig ] we report the residual energy results for the case of the asymmetric ( top ) and symmetric potential ( bottom ) .
the results are obtained using @xmath218 trotter s slices and @xmath197 bisection steps .
averages and statistical errors are calculated , as usual , with @xmath188 repetitions of every annealing run .
it is clearly seen that , as in the classical case , the lorentzian move
_ alias _ , in the present context , the _ relativistic kinetic energy _ greatly accelerates the qa behavior for the difficult asymmetric potential case . as usual ,
the `` simpler '' symmetric potential case does not show a qualitative change . , but for a qa based on the relativistic kinetic energy of eq .
[ relativistic_hamiltonian : eqn ] , implemented via a bisection algorithm adapted to lorentzian moves .
as a reference , the results obtained by exact integration of the imaginary - time schrdinger and by the previous gaussian - based pimc - qa with and without instanton move are still reported.,width=302 ]
summarizing , we would like to briefly stress some of the major conclusions reached in this paper . 1 .
_ role of move choice in classical annealing . _ : : the choice of the monte carlo ( mc ) moves strongly influences the spectral properties of the mc markov transition matrix , thus modifying in a strong way the annealing properties . in some sense , the _ effective landscape _ of the problem is really determined not only by the actual potential landscape , but also by the choses markov transition matrix , the latter determining the _ neighborhood _ of a given configuration @xcite . in our double - well example a metropolis mc with lorentzian - distributed proposed moves ended up showing the best annealing performances , overperforming not only the other classical mc choices , but also any path - integral monte carlo ( pimc ) algorithm which we were able to implement .
2 . _ tunneling and sampling difficulties of pimc . _ : : a tunneling event , and its associated landau - zener crossing , can cause severe difficulties to a `` state - of - the - art '' pimc - qa algorithm .
the difficulty is associated to a poor sampling of the action which misses the rare , but all - important , tunneling events .
this problem can be cured by the ad - hoc introduction of `` instanton moves '' , but this cure , while instructive , uses detailed information on the potential landscape that is generally not available .
the generalization of these `` instanton moves '' to more complicated potentials , let alone to generic hard optimization problem , seems basically impossible , for it would require , among other things , a knowledge of the location of the minima were tunneling occurs .
_ other limitations of pimc . _ : : pimc - qa suffers from a certain number of other limitations . first of all , it works with a _ finite temperature _ @xmath52 , and this sets up a thermal energy lower bound below which we can not possibly go ( this limit was particularly clear in our double - well example ) .
second , a large number of trotter slices @xmath159 can cause additional sampling problems in that an effective uncorrelation of the configurations becomes harder and harder , even if a multi - step bisection algorithm is employed .
moreover , the _ trotter break - up _ itself , see eq .
[ trotter_theorem : eqn ] , can cause difficulties whenever , for a generic kinetic energy @xmath52 , the form of the free propagator , generalizing the simple gaussian result of eq . [ hkin_primitive : eqn ] , is not known analytically .
this was indeed the difficulty met in a pimc - qa study of the traveling salesman problem @xcite .
_ role of kinetic energy in qa . _ : : the choice of the kinetic energy is clearly all important in qa : sec .
[ results3:subsec ] , illustrating the improvements in pimc - qa upon using a relativistic kinetic energy , is particularly instructive . in our one - dimensional example , this choice corresponds to a free propagator of lorentzian form , which we implemented by an appropriately modified bisection scheme ( see app . [
lorentzian_move : sec ] ) .
the convenience of such a scheme in more general instances is not _ a priori _ obvious and further tests on higher - dimensional problems would be needed . in view of the previous points ,
it is fair to say that exploring alternative quantum monte carlo schemes for performing stochastic implementations of quantum annealing remains an important and timely issue for future studies in this field .
one such scheme , which in principle does not suffer from many of the limitations of pimc , is the green s function monte carlo scheme .
test applications of a gfmc - qa strategy to the ising spin glass are currently under way and will be reported shortly @xcite .
this project was sponsored by miur through firb rbau017s8r004 , firb rbau01lx5h , prin / cofin2003 and 2004 , and by infm ( `` iniziativa trasversale calcolo parallelo '' ) .
we acknowledge illuminating discussions with demian battaglia .
one of the author ( ls ) is grateful to llorenc brulla for his generous help in discussing details about the pimc algorithm .
in this appendix we shall use analytical tools to explain some of the most relevant features of the markov transition matrix spectra presented in sec .
[ spectral : subsec ] .
the question we want to address has to do with the low - lying spectrum of a master equation of the form @xmath219 where @xmath85 is the transition operator associated to a certain choice of trial move in a metropolis mc scheme . as we have seen previously
( see sec .
[ spectral : subsec ] ) , the box proposed move case admits a simple geometrical interpretation . for the general case ,
which is much harder to treat , we shall sketch here the main steps of a more systematic procedure used to extract information on the gap of @xmath80 , referring the reader to ref . for the many technical details . as explained in sec .
[ spectral : subsec ] , it is possible to find a complete set of eigenvectors and eigenvalues of the matrix @xmath63 , defined through the short - time expansion of this equation : @xmath65 .
@xcite in particular , one can find a linear transformation from the ( generally ) non - symmetric operator @xmath66 to a symmetric one , @xmath220 , which preserves its spectral properties @xcite , and is easier to diagonalize in a standard way .
moreover , one can apply the same linear transformation to the original transition matrix , @xmath80 ; this leads to the expansion of the master equation solution reported in eq .
[ spectral_decomposition : eqn ] , having taken the appropriate time - continuum limit . however , in order to find the @xmath80 eigenvectors and eigenvalues ( and in particular its spectral gap , @xmath84 , defined in sec .
[ spectral : subsec ] ) , it is better to keep @xmath32 finite and consider its simmetrized version , that turns out to be the exponential operator @xmath221 , where the appropriate @xmath222 is uniquely determined by the move range @xmath38 .
we notice that the spectrum of @xmath66 , @xmath223 , introduced in sec .
[ spectral : subsec ] is simply related to the spectrum of the original transition matrix @xmath80 , @xmath224 , by the equation @xmath225 . in the following
, we shall refer only to the gap of @xmath80 .
the potential @xmath0 will be always assumed to have the usual double - well form . at very small temperatures , the equilibrium distribution @xmath140 is concentrated around the local minima at @xmath226 @xcite .
therefore , there is definitely a regime of parameters where a two - level system approximation for @xmath80 must hold . by exploiting this approximation
, one can show ( see ref . )
that the spectral gap @xmath227 of the transition operator @xmath80 can be expressed , at small temperature @xmath52 , in terms of a well - to - well propagator of the evolution operator of @xmath220 as follows : @xmath228 where @xmath229 .
the crucial quantity we need , therefore , is the well - to - well propagator @xmath230 .
an expression for this propagator can be obtained by performing a trotter decomposition and writing a feynman path - integral @xcite .
the final result can be cast by an appropriate saddle - point approximation in the form : @xmath231 where @xmath232 is the function introduced in sec .
[ spectral : subsec ] ( see fig . [
fig_double_well_barrier : fig ] ) , and @xmath233 accounts for the gaussian fluctuations around the saddle - point classical solution of the feynman integral , which do not modify the important exponentially activated behavior provided by the term @xmath234 .
@xmath235 , finally , is just the proposal move distribution : @xmath236 ( see fig .
[ fig_double_well_barrier : fig ] ) .
this rather simple equation for the well - to - well propagator can be further treated by another standard saddle - point approximation of the relevant one - dimensional integral .
we shall discuss here the cases we are interested in : box , gaussian , and lorentzian . *
box move * : @xmath159 was defined in eq . [ p_transition_matrix_box : eqn ] .
we recall that the function @xmath232 , introduced in sec .
[ spectral : subsec ] , is a monotonically decreasing function of its argument , which exhibits an infinite first order derivative for @xmath237 .
there is no true saddle - point for such a move , since the functions involved in the exponential part of eq .
[ transition_amplitude_final : eqn ] are all monotonic . on the other hand , in the case of the box move the extremes of integration in eq .
[ transition_amplitude_final : eqn ] can be taken as @xmath238 $ ] instead of @xmath239 $ ] , since its transition matrix , @xmath235 , is zero outside the former interval .
therefore , the largest contribution to the whole integral is due either to @xmath237 or @xmath240 , which are the two integration extremes .
it turns out that the latter choice guarantees the largest exponential , and is therefore the right one . as a consequence
, we can write down that : @xmath241 this transition amplitude is arrhenius - like , and so does the gap function ( see eq .
[ gap_value_integral : eqn ] ) which reads : @xmath242 where @xmath243 , as anticipated in sec .
[ spectral : subsec ] from geometrical considerations .
this is indeed the behavior seen in fig .
[ box_vs_t : fig ] . * gaussian move * : @xmath159 was defined in eq .
[ p_transition_matrix_gau : eqn ] . in this case , a true saddle - point is possible if the following equation has a solution : @xmath244 since @xmath245 and @xmath246 , then , in the limit @xmath247 we have that either @xmath237 or @xmath248 are solutions of eq . [ saddle_point_gau : eqn ] .
the second choice guarantees the largest contribution to the integral in eq .
[ transition_amplitude_final : eqn ] , and must be therefore selected .
the resulting expression for the transition amplitude is : @xmath249 and the corresponding gap value is : @xmath250 fig .
[ all_vs_sigma_tot : fig ] shows that such a dependence perfectly fits the simulation data .
we emphasize that even the coefficient within the exponential part of the fitting function matches the theory ( we recall that @xmath251 ) : only an overall prefactor is used as fitting parameter . as observed in sec .
[ spectral : subsec ] , the previous equation will be no longer valid for very small values of the temperature @xmath52 , where an ( avoided ) crossing between the first eigenvalue @xmath252 and the second eigenvalue @xmath253 will occur . on the other hand , for large values of @xmath38 we see from fig .
[ all_vs_sigma_tot : fig ] that the data deviate from the behavior predicted by eq .
[ theory_gap_gau : eqn ] .
this is the consequence of another level crossing . *
lorentzian move * : @xmath159 was defined in eq . [ p_transition_matrix_lor : eqn ] . to cut a long story short ( calculations
proceed similarly to the box and gaussian case ) , we write down the result . the gap value is : @xmath254 and also in this case the model agrees with the data , see fig .
[ all_vs_sigma_tot : fig ] , within a multiplicative fitting coefficient .
in this appendix we want to show how to approach , within a path - integral monte carlo , a hamiltonian with _ relativistic _ kinetic energy : @xmath255 the kinetic operator @xmath256 is singular in a real space representation , but the corresponding density matrix operator can be written in a simple closed form : @xmath257 this is a _ lorentzian ( or cauchy ) distribution _ , while the kinetic density matrix operator considered in sec .
[ pimc : sec ] , eq .
[ hkin_primitive : eqn ] , was a gaussian one .
it is possible to obtain a generalization of the lvy construction @xcite for this kind of lorentzian move , observing that both gaussian and lorentzian distribution are stable under convolution ( a property shared by all the so - called _ lvy distribution _
the main idea is to obtain a closed form of the constrained kinetic operator : @xmath258 which is the probability of moving the @xmath55-th bead to @xmath259 , keeping fixed the @xmath260-th and the @xmath261-th beads in @xmath173 and @xmath262 , respectively .
we recall that this object is the building block of the bisection algorithm @xcite . in the lorentzian case
, it reads : @xmath263\ , \left[\left(\frac{x'_i - x_{i-1}}{\gamma\,\delta\,t}\right)^2 + 1\right]}\;.\ ] ] unfortunately , this is no longer a lorentzian distribution , and therefore it is not trivial to sample it ( for instance , by making use of the usual transformation technique ; see ref . ) .
however , after some algebra , and a smart change of variables ( taking due care of the jacobian of such a transformation , which changes @xmath52 onto @xmath264 ) , we obtain a simpler form : @xmath265\,\left[\left(y - a\right)^2 + 1\right]}\;,\ ] ] where @xmath266 , @xmath267 , and @xmath268 finally , with a bit of more algebra , we find that : and also that @xmath270 .
an efficient strategy to sample @xmath271 is , therefore , to sample @xmath272 and then to make use of the usual rejection technique ( see ref . ) .
the primitive of @xmath272 can be easily computed .
it reads : @xmath273 inverting @xmath274 , and making use of an equidistributed random number @xmath275 $ ] , we find that : @xmath276 \\ -d + \sqrt{d^2 + ( a^2 + 1 ) } & \mbox{for } \ ; u\in({1\over2},1 ] \end{array } \right.\ ] ] where @xmath277\;.\ ] ] provided @xmath278 $ ] another equidistributed random number we shall accept the @xmath279 obtained above according to the condition : @xmath280 it results from standard considerations ( see ref . )
that the average acceptance of this method will be around @xmath281 .
( this means that in order to generate @xmath53 random numbers distributed according to @xmath282 , one has to try on average @xmath283 times . ) finally , the original middle - point value @xmath259 appearing in @xmath282 is obtained by inverting the original change of variables in eq .
( [ change : eqn ] ) , @xmath284 this sampling method is a bit cumbersome , but it can be easily implemented as a computer algorithm .
its generalization , namely @xmath285 , which represents the probability of moving the @xmath55-th bead to @xmath259 having fixed the @xmath286-th and the @xmath287-th beads in @xmath288 and @xmath289 , can be easily derived ( it is enough to change @xmath32 with @xmath290 ) .
this is what one needs in order to implement the bisection scheme @xcite . in this way ,
a pimc algorithm using such a bisection sampling scheme implements correctly a _ relativistic _ choice of hamiltonian as in eq .
[ hrel : eqn ] .
moreover , since @xmath291=0 $ ] , for such a relativistic - pimc the simple primitive approximation is already exact to fourth - order @xcite .
we finally report the virial - centroid estimators for the kinetic an potential energy for the relativistic - pimc case : @xmath292 where @xmath293 is the centroid coordinate .
the way to derive them is completely similar to the non - relativistic case considered in refs . . | we present results for a variety of monte carlo annealing approaches , both classical and quantum , benchmarked against one another for the textbook optimization exercise of a simple one - dimensional double - well . in classical ( thermal ) annealing ,
the dependence upon the move chosen in a metropolis scheme is studied and correlated with the spectrum of the associated markov transition matrix . in quantum annealing
, the path - integral monte carlo approach is found to yield non - trivial sampling difficulties associated with the tunneling between the two wells .
the choice of fictitious quantum kinetic energy is also addressed .
we find that a `` relativistic '' kinetic energy form , leading to a higher probability of long real space jumps , can be considerably more effective than the standard one . |
one of the most common tasks in microarray analysis is to identify a list of genes that are differentially expressed under two conditions , such as being affected by a disease vs. normal , before vs. after a medical treatment , and one vs. another disease subtype .
the number of genes on the top - ranking list is usually much smaller than the total number of genes on the chip , @xmath2 .
if the same type of microarray chip is used for two different studies ( e.g. disease - a vs. control , and disease - b vs. control ) , two differentially expressed gene lists can be obtained , with @xmath3 and @xmath4 genes .
researchers often find the same genes appear in both lists and hypothesize that these common genes are involved the etiology of both diseases .
however , for such a hypothesis to be convincing , one has to first estimate the probability for overlapping genes by chance alone . in other words ,
if two lists of genes are selected out of @xmath2 genes randomly , we would like to calculate the probability for @xmath5 genes in common in the two lists , with the lengths of the two lists being @xmath3 and @xmath4 .
this overlapping probability is known to follow the hypergeometric distribution .
the name hypergeometric distribution was first used in @xcite , and was popularized by its role in fisher s exact test @xcite . in microarray analysis ,
overlapping probability and hypergeometric distribution mainly appear in testing the enrichment of genes in certain functional category @xcite . in this application ,
the first list is the top - ranking differentially expressed genes , and a gene selection process is involved .
the second list is nevertheless given : @xmath4 genes are known to be in a pathway , a member of a protein family , described by a gene ontology term , etc .
one asks the question on chance probability for @xmath5 out of @xmath3 selected genes to be in a given pathway , a protein family , and describable by a gene ontology term .
fixing @xmath4 or not is the main difference between their application and ours . when a different gene selection criterion is used , the number of genes in the two top - ranking lists of two studies ( @xmath3 and @xmath4 ) will also change . because the stringency of a gene selection criterion is always adjustable and to some extent arbitrary
, we would like to examine whether these changes will affect the overlapping probability . at two extreme situations , very small @xmath6 and very large @xmath7 ,
it is clear that the number of overlapping genes is @xmath8 and @xmath9 .
these @xmath5 values appear 100% of the times , so the corresponding @xmath0-value is equal to 1 , i.e. , not significant . for intermediate @xmath10 values , it is not clear what the overlapping probability and significance will be , and it is the topic of this abstract .
given integers @xmath2 , @xmath3 , @xmath4 , @xmath5 ( @xmath11 and @xmath12 ) ) , the hypergeometric distribution is defined as @xmath13 where @xmath14 is the number of possibilities of choosing @xmath5 objects out of @xmath2 objects : @xmath15 $ ] . when @xmath3 genes are randomly chosen from the total of @xmath2 genes , and another random sampling leads to @xmath4 genes , the probability that the two lists of genes have @xmath5 in common is exactly the hypergeometric probability @xmath16 .
this can be proven by the following steps : 1 ) the total number of possible choices for the two lists of genes is @xmath17 .
2 ) there are @xmath18 possibilities for choosing the first list .
3 ) among the @xmath3 genes in the first list , there are @xmath19 possibilities for choosing @xmath5 genes to be in common with the second list .
4 ) in the second list , besides the @xmath5 genes that are in common with the first list , the remaining @xmath20 genes are chosen among the @xmath21 leftover " genes not in the first list , thus @xmath22 possibilities . the @xmath16 is simply ( # 2 @xmath23 # 3 @xmath23 # 4 ) / # 1 . note
that @xmath3 and @xmath4 can be switched without changing the @xmath16 value .
it is usually more interesting to calculate the sum of @xmath16 for @xmath5 s equal or larger than the observed value ( i.e. , the @xmath0-value ) : @xmath24 in statistical package @xmath25 ( _ http://www.r-project.org/_ ) , there are at least two ways to calculate the overlapping @xmath0-value .
the first is to use the accumulative distribution of hypergeometric distribution , _
phyper(m , @xmath3 , @xmath21 , @xmath4 ) _ : @xmath0-value @xmath26 if @xmath27 , and @xmath0-value=1 if @xmath8 .
the second method is to use the @xmath0-value from the fisher s exact test on the following 2-by-2 table : @xmath28 the two approaches lead to the identical result . -90
in hypergeometric distribution , the number of overlapping elements @xmath5 is an independent variable from the the list lengths @xmath29 . in order to get a rough idea on how @xmath5 changes with the list lengths , we use three real microarray datasets .
theese studies concern three autoimmune diseases : rheumatoid arthritis ( ra ) , systemic lupus erythematosus ( sle ) , and psoriatic arthritis ( psa ) , described in details in @xcite .
the number of controls ( c ) and patients ( p ) in these three datasets are ( c=39 , p=46 ) , ( c=41 , p=81 ) , and ( c=19 , p=19 ) , respectively .
the total number of genes / probe - sets is @xmath3022283 , and the expression levels are log transformed .
genes are ranked for their degree of differential expression which can be measured by various tests or models , such as @xmath1-test and logistic regression . for any pair of studies , with a fixed number of top - ranking gene lists @xmath31
, one can count the number of overlapping genes @xmath5 and the proportion @xmath32 .
fig.[fig1 ] ( left column ) shows this proportion as a function of @xmath31 for three study - pairs ( ra - sle , sle - psa , ra - psa ) as well as for two ranking methods ( @xmath1-test and logistic regression ) .
similar overlapping proportion of two random shuffled lists is also indicated in fig.[fig1 ] as crosses . when @xmath31 is small , @xmath5 is more likely to be zero , so the proportion is also zero .
when @xmath31 approaches the total number of genes , @xmath2 , all genes are overlapping genes , and the proportion is 1 . fig .
[ fig1 ] indeed shows these trends at the two extreme points . in order to check behavior in - between ,
we draw a reference line in fig.[fig1 ] ( left column ) that assume a linear relationship between @xmath33 and @xmath34 .
most of the points on fig.[fig1 ] are above this line , and the overlapping proportion of two random lists is exactly on this line .
to have an idea of the absolute number of common genes more than expected by random chance , fig.[fig1 ] ( right column ) plots the observed @xmath5 subtract the expected @xmath35 as a function of @xmath31 . the maximum difference between the observed and expected
is reached between @xmath36 and @xmath37 .
the difference of observed and expected @xmath5 s can be as much as 600800 . -90
the overlapping @xmath0-value corresponding to the @xmath5 counts plotted in fig.[fig1 ] was calculated by the hypergeometric distribution , and is shown in fig.[fig2 ] : @xmath38-axis is @xmath39-value ) , and @xmath40-axis is @xmath31 .
six lines are shown for three comparisons ( ra - sle , sle - psa , ra - psa ) and two measurements of the differential expression ( @xmath1-test and logistic regression ) .
zero @xmath0-values are converted to 2.2 @xmath41 which is the minimum value reported by @xmath25 program .
fig.[fig2 ] shows that besides the two ends ( @xmath42 and @xmath43 ) where the @xmath0-value is 1 , the overlapping significance quickly increases with the length of top - ranking gene list @xmath44 ) , and can be extremely significant when a large number of genes are kept in the two lists for comparison .
this result confirm our previous suspicion that overlapping significance is a function of the gene list lengths .
if the selection of @xmath29 is arbitrary , the overlapping significance thus calculated is also arbitrary .
it is not surprising that overlapping significance may keep increasing ( or , @xmath0-value decreasing ) with the increase of @xmath31 , because @xmath0-value in general depends on the sample size .
when a signal is real ( true positive ) , @xmath0-value will monotonically decrease with the sample size . on the contrast , if a true signal is absent , the sample size does not affect the conclusion . as can be seen in fig.[fig2 ] , the overlapping significance for two random lists does not really change with @xmath31 . one may argue that it is unlikely to consider top 5000 genes as being differentially expressed , because by a typical selection criterion ( e.g. @xmath0-value of @xmath1-test smaller than 0.01 , with or without multiple testing correction ) , the number of genes selected is less than a few hundreds . however , as can be seen in fig.[fig2 ] , even in the range of 10500 , the overlapping @xmath0-value changes dramatically .
this pitfall of gene - list - length dependence of overlapping @xmath0-values has not been noticed before perhaps because in other application of hypergeometric distribution for calculating overlapping probability , the length of the second list @xmath4 is fixed , for example , in the study of overrepresentation of genes in certain pathway .
the number of overlapping genes @xmath5 is then constrained from above by @xmath45 even though the length of the first list , @xmath3 , might increase by relaxing the gene selection criterion . -90
there are many genes / probe - sets on the microarray chip that do not register much signal .
since these low - expressed genes are lowly expressed in both control and patient samples , they usually do not appear in the top - ranking differentially expressed gene list .
fig.[fig3 ] shows @xmath39-value ) of each gene of 3 @xmath1-tests sorted by average expression ( log - transformed ) across all 245 samples in 3 datasets ( for both cases and controls ) .
although we can not use the average expression level to predict the degree of differential expression , there is a general trend for low - expressed genes to rank lower in the differentially expressed list as seen from fig.[fig3 ] .
we removed 7000 genes with lower overall expression across all samples , leaving @xmath46 genes .
figs.[fig1 ] and [ fig2 ] are reproduced in fig.[fig4 ] for the dataset with a reduced gene pool . as in figs.[fig1 ] and
[ fig2 ] , the observed number of overlapping genes @xmath5 is much larger than the expected , though the difference peaks at 400600 , as versus 600 - 800 in fig.[fig1 ] .
the overlapping significance as measured by @xmath47-value ) again quickly moves up with @xmath31 as shown in the last column of fig.[fig4 ] .
the qualitative similarity between figs.[fig1 ] , [ fig2 ] and fig.[fig4 ] indicates that the presence of low - expressed genes does not affect our conclusion .
using the hypergeometric distribution to calculate the overlapping probability between two top - ranking differentially expressed genes in two studies , we have shown that the overlapping significance depends on the stringency of gene selection criterion , or equivalently , the length of the gene lists .
this observation presents a problem when an overlapping @xmath0-value is reported but the gene selection criterion is not specified . on the other hand , the increase of the overlapping significance with the gene list length can be an indication that the significant overlapping of genes is a true signal .
the overlapping probability calculated here assumes the two top - ranking gene lists are selected from the same pool of @xmath2 genes .
if the two studies are based on different chip platforms , the two initial gene pools are not identical , though there are perhaps certain common genes .
we plan to derive the overlapping distribution for this situation .
we also plan to study the probability for genes appearing in three top - ranking gene lists .
although a permutation based approach comparing multiple studies was proposed in @xcite , there is no analytic formula available .
we would like to thank prof .
richard friedberg for suggestions .
g. finocchiaro , f. mancuso , h. muller , mining published lists of cancer related microarray experiments : identification of a gene expression signature having a critical role in cell - cycle control " , _ bmc bioinf .
_ , vol 6(suppl 4 ) , 2003 , s14 . l. tian , s.a .
greenberg , s.w .
kong , j. altschuler , i.s .
kohane , p.j .
park , discovering statistically significant pathways in expression profiling studies " , _ proc .
_ , vol 102 , 2005 , pp 13544 - 13549
batliwalla , e.c .
baechler , x. xiao , w. li , s. balasubramaniuan , h. khalili , a. damle , w.a .
ortmann , a. perrone , a.b .
kantor , m. kern , p.s .
gulko , m. kern , r. furie , t.w .
behrens , p.k .
gregersen , peripheral blood gene expression profiling in rheumatoid arthritis " , _ gene and immunity _ , vol 6 , 2005 , pp 388 - 397 .
baechler , f.m .
batliwalla , g. karypis , p.m. gaffney , w.a .
ortmann , k.j .
espe , k.b .
shark , w.j .
grande , k.m .
hughes , v. kapur , p.k .
gregersen , t.w .
behrens , interferon - inducible gene expression signature in peripheral blood cells of patients with severe lupus " , _ proc .
_ , vol 100 , 2003 , pp 2610 - 2615 .
batliwalla , w. li , c.t .
ritchlin , x. xiao , m. brenner , t. laragione , t. shao , r. durham , s. kemshetti , e. schwarz , r. coe , m. kern , e.c .
baechler , t.w .
behrens , p.k .
gregersen , p.k .
gulko , microarray analyses of peripheral blood cells identifies unique expression signature in psoriatic arthritis " , _ mol
_ , 2006 , to appear .
rhodes , j. yu , k. shanker , n. deshpande , r. varambally , d. ghosh , t. barrette , a. pandey , a.m. chinnaiyan , large - scale meta - analysis of cancer microarray data identifies common transcriptional profiles of neoplastic transformation and progression " , _ proc .
_ , vol 101 , 2004 , pp 9309 - 9314 . | when the same set of genes appear in two top ranking gene lists in two different studies , it is often of interest to estimate the probability for this being a chance event .
this overlapping probability is well known to follow the hypergeometric distribution .
usually , the lengths of top - ranking gene lists are assumed to be fixed , by using a pre - set criterion on , e.g. , @xmath0-value for the @xmath1-test .
we investigate how overlapping probability changes with the gene selection criterion , or simply , with the length of the top - ranking gene lists .
it is concluded that overlapping probability is indeed a function of the gene list length , and its statistical significance should be quoted in the context of gene selection criterion . |
we propose a network that unifies computer - readable knowledge , rules , inference , reasoning , and more equivalent aspects .
the network evolves with time toward gathering more meaningfull information .
humanity is not wild , but engineered .
it is , and always was , engineered and manipulated by relatively small groups of people .
our media , education , economy , law , politics , ethics , are all shaped mainly according to the views of small groups .
those groups tell us we are englightened , but in fact , the situation is very much of opposite nature .
immanuel kant began his famous essay `` what is enlightment ? '' , and he defines : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ `` enlightenment is man s emergence from his self - imposed nonage .
nonage is the inability to use one s own understanding without another s guidance .
this nonage is self - imposed if its cause lies not in lack of understanding but in indecision and lack of courage to use one s own mind without another s guidance .
dare to know !
( sapere aude . )
have the courage to use your own understanding , is therefore the motto of the enlightenment . ''
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ those 1784 ( 1984 ? ) words are in fact relevant for today more than ever : today we can break the chains of opression and let our own voice emerge .
about a century later , franz kafka described the inability of the individual to stand against irrational and moraless bureaucratic systems .
while the laws are never actually given to the citizens , they still must obey them or otherwise suffer the wrath of society acting in violence against them , while the system contradicts itself : on one hand they claim they can never define law in a closed form , and on the other hand , they always find a way to justify their actions according to the law .
today , they also blame their computers and the records or decisions it allows to keep or modify .
so maybe law can not be formalized for us , but can be formalized for them ?
law can and should be formalized .
the most important property a law system should have is a consistent ethical basis ( e.g. constitution ) and consistent implications from this basis to laws themselves .
but who should formalize the laws ? formalizing laws will pose even a greater danger if done by centralized hands .
it is evident that democracy is incapable of assisting : the way in which voting is done today is far from assuring a consistent , moral , and functioning system . until recent times
, this was indeed an inevitable situation .
now we have mathematical and technological ways to create laws collaboratively , while preserving on frames we set for ourselves , like consistency , votes , or minimal requirements .
centralization of law is only one thing .
what about centralization of information ?
think of the following situation : google has the most valuable information in the world , namely , `` what people want '' .
they do not give us access to their databases . but imagine , what if we could access their data : we would not only obtain this important information , but also have the ability to perform a much more sophisticated search .
we could , for example , query about topics related to a given topic , or automatically build new aggregated data from the database , and basically have endless additional uses .
communication is also centralized .
when communicating by electronic means , what we say is often intercepted by unintended parties .
our privacy is deeply vulnerable .
we do not tend to have our own website and post our thoughts there , but we do it on centralized locations like facebook .
we do it because centralized hands give us better technology , with long term support and less bugs .
we pay for it with our privacy , endless ads and marketing junk , and we even let them manipulate who our friends will be .
can we keep all benefits and have a moral high - end software ?
@xmath0-chain s goal is to unite , yet keeping decentralization , humanity s knowledge and thinking , know - hows and communication , laws and opinions , all into one giant shared database that is able to be coherent and consistent , to be queried meaningfully , to reuse information / code / data efficiently , to allow all kind of social operations and communications to be done with no unwanted guests , to allow every user set their own specific rules and to be part of communities sharing the same thoughts , or goals , or needs .
@xmath0-chain does not enforce anyone to subscribe or follow one or another ruleset .
but it lets all mental and technological benefits to be combined for the sake of better humanity .
obviously , not all questions are solvable , some better solved by humans and some better solved by machines , and some are solvable with many machines and a lot of knowledge .
once one is able to solve a puzzle , namely : to prove a given claim , it gets into the network . until then
, it stays out .
this vision is not new , and probably began with the semantic web as we will describe on section 5 , but could not be fulfilled without the nature of decentralization , which also gives the ability to fairly incentivize participants .
moreover , this vision in its decentralized version is not new as well , but no serious attempts were made before to make it happen .
it is about time to begin doing to many issues bothering humanity what satoshi nakamoto began to do to the monetary system .
it should be noted that @xmath0-chain is nt a coin as for itself , and will bootstrap as a decentralized network , where developers are just like everyone , without any kind of so - called premine or any currency in hand , since the system itself has to be bootstrapped practically without rules .
* db : database * p2p : peer - to - peer * prop[s ] : proposition[s ] - a claim or definition to be understood under strict logic * dht : distributed hash tables * dtlc : dependently typed lambda calculus * tfpl : totally functional programming language * rdf : resource description framework * n3 : notation3 language
we present @xmath0-chain , a fully decentralized p2p network being a generalization of many centralized and decentralized p2p networks , including the blockchain .
we aim to generalize the concept as much as we find , and give users the ability to implement virtually any p2p network over @xmath0-chain .
its interpretations , uses , and consequences are far from being a p2p network only , and include software development , legal , gaming , mathematics and sciences , logic , crypto - economies , social networks , rule - making , democracy and votes , software repositories ( like decentralized github+appstore / google play ) , decentralized storage , software approval and verification , even `` doing your homework in history or math '' in some sense ( stronger sense that search engines ) , and many more aspects - chain .
bitagoras will work over it as well . ] .
@xmath0-chain can be defined by several equivalent definitions : * a shared db of rules , with a client that is able to change the rules , obey the rules , infer new rules from given rules , and make sure rules are consistent . *
a shared db of computer programs code being collaboratively composed with revision control and custom permissions ( like git ) , with a client that is able to run code , reuse code from existing programs , and verify programs against formal specifications which are programs themselves . * a shared db of props , with a client that is able to state new props , prove props with custom derivations rules which are props themselves , and verify proofs to given props . * a shared db of ontologies , which are definitions of types ( taxonomies ) and their relations , with a client that is able to propose new ontologies , make sure that ontologies are consistent , and query the ontologies db various queries .
* a decentralized nomic game .
-chains ] at its basis , @xmath0-chain is nothing but a db of quads being tuples of four words each , namely : context , subject , predicate , object . on this note
we show how rules , programs , proofs , and ontologies are all naturally representable by a unified way , and point to some of the far - reaching uses of such a system .
every node contains knowledge , at the very intelligent sense , namely the ability to make inference and implication using logic .
they do not aim to represent a human intelligence , but their knowledge is first and foremost how to communicate and transfar liable information .
technically , this is can be done by formalizing with ontologies algorithms like dht and blockchain .
but when looking at this in a unfied way as virtual creatures being able to ask rather retrieve , know rather store , talk rather comminucate , consider rather verify , having only mouth and ears and ontologies and a reasoner .
we give an informal explanation of rules as programs as proofs . for a formal derivations , we refer the reader to the literature of type theory . by intuition , running a program can easily be seen as equivalent to obeying the rules , while the rules are the code itself .
the non - triviality lies on the other direction : can all rules be formalized as programs ?
the answer is positive , in a sense , and we will explain it through the proofs interpretation . from abstract logic point of view ,
a proof is a path taken from given hypotheses and axioms , and ends up with a result to be proved under the hypotheses , while the path has to be taken according to given derivation rules , and follows the notion of ( @xmath1 ) .
following are examples of derivation rules : * modus ponens : if a implies b and a is true , then b is true : @xmath2 * cryptographic signature : it is impossible to generate a valid signature ( up to assumed validation procedure ) without access to the corresponding private keys . *
a mapping from data to its hash is one - to - one and noninvertible . obviously , this is false under pure mathematical logic , but it is valid in cryptography being a practical science . *
hilbert s formal inference rules and gentzen s formal inference rules .
it should be noted that derivation rules can be stated as axioms , and axioms as hypotheses .
hence , a proof is inferring a statement from other statements , or equivalently , deriving a rule given other rules . looking at definitions ( taxonomy ) as rules themselves , we see that rules are proofs and proofs are rules , and that proving is inferring . this non - trivial result is the celebrated curry - howard isomorphism .
as there are many kinds of programs and many kinds of proofs , we are interested in a space where programs always halt . roughly speaking , such programming language fall under the class of tfpl , e.g. idris , agda , coq , which correspond to dtlc .
such and other class are isomorphic to the classes of proofs .
tfpl are not turing complete .
this lets us escape from many paradoxes arising from turing machines formalism , promises us that programs always halt , and give us strong abilities to claim and prove claims about the program . by this
, we can prove that a given code satisfies a given unit - test .
we can also prove the execution of the program .
it should be noted that practiaclly , not being turing complete at this sense gives only advantages : any application one can come up with does not require turing completeness , but dtlc is enough .
turing completene languages can do things that dtlc can not only at extreme ( mostly infinite ) theoretical cases .
logical proposition can be interpreted as rules , and vice versa .
proofs from axioms can be interpreted as inferring rules from other rules , and vice versa .
computer programs written in a totally functional programming language can be interpreted as constructive proofs , and vice versa .
it should be stated that the correspondence between proofs and computer programs is so strong , that it gives rise to new foundation of mathematics , rather cantors / zfc set theory , namely : category theory .
it turns out that proofs and programs are isomorphic to category theory , at some sense .
this is the curry - howard - lambek isomorphism , aka computational trinitarianism .
this is really a long story , and we re going to make it very relatively short .
it all began when the inventor of the world wide web , tim berners - lee , came up with a vision that all data online will be machine readable , and build a web of relations between objects and types . that s the semantic web vision in a nutshell .
for this goal , a vast amount of tools has been developed and still being developed .
nowadays there s a huge and growing amount of ontologies from all aspects of life : basic web components and logical entities and relations ( owl ) , legal ( cf .
lkif ontologies ) , software , medicine , e - commerce , cyptography , social media , geography , news , and many more .. another notable project is _ dbpedia _ , that already formalized millions of concepts from wikipedia with consistent ontologies .
there is also a vast amount of tools to manipulate rdf data , a list can be found on the semantic web wiki .
the same growth can be observed in mathematics and other sciences formalized for coq , all naturally translatable to rdf , and containing thoushands of mathematical theorems with their full proofs .
the basic representation of concepts is standardized by rdf ( resource description framework ) .
rdf is a language expressing ontologies .
an ontology contains definitions of types ( taxonomy ) like `` dog is an animal '' , and relations between them like `` all dogs has four legs '' .
of course , a definition of four has to make sense as well , like the definition of `` has '' .
notation3 ( n3 ) is a language makes it more convenient and human - readable to write rdf ontologies .
it is logically powerful enough to represent dtlc .
moreover , it can always be converted into quads and vice versa , which is the natural format for @xmath0-chains , as mentioned at the end of the introduction .
rdf being exactly pure logic can also be converted into english , in some limited sense , and vice versa .
it is more machine - readable english than human - readable , but still very natural .
we of course speak about the attempto project .
eulergui is a veteran ide for various reasoning engines , demonstrating the power of ontologies .
some of its supported formats are rdf , rdfs , n3 , owl , uml , sparql , attempto ( controlled english ) , and even java jars .
it can output java code that builds a ui according to an ontology , using rule based swing .
it supports four different and powerful reasoning engines , logican queries , explained proofs , consistency checks , fuzzy logic , graphical visualization , and more .
it can be integrated with the deductions engine which is itself written in ontologies .
it is described on their website as `` artificial intelligence techniques applied to common software tasks , using first order logic through n3 + owl ontologies and rules '' .
both eulergui and deductions are compatible with the powerful reasoning engine drools . on drools website one can find pros and cons of ontologies based development .
the reader is invited to get introduced to this world from the linked materials .
a more recent tool is protege , a professional open - source ontologies ide .
given we formalize our rules as an ontology .
what are we going to do with them ?
we take an example tool called cwm .
we can ask it questions that are answerable from within the rules , even if not explicitly but by inference . we can also verify the consistency of rules .
the reader is invited to take a look at its various tools , including the reasoner and the owl verbalizer ( owl is rdf with some basic ontologies defined , and is a w3c standard ) . see appendix for demonstrative screenshots .
dtlc based languages can implement almost everything , and in practical sense - everything .
while turing complete languages suffer from undecidability , on dtlc languages we can look at the code as logic we can work with and prove various useful claims directly from the code , like that the program does not use the internet , or accesses only certain files , , show it halts , prove its execution path etc . the transition from dtlc to rdf is nt trivial but two examples of how it can be done can be found at @xcite .
the structure of the program ( almost ) does nt change and namings can be preserved .
the rdf format can be kept for being human readable , but for machine - proofs we do not stay on the rdf representation but use smt solvers for reasoning .
as stated above , @xmath0-chain can generalize any p2p network .
a decentralized p2p client is a state machine that decides what to do given various inputs , according to some rules .
dht is an example of a p2p architecture .
one of the products using it is bittorrent .
the most common flavor of dht is kademlia .
bitcoin s network is dht itself , but with full replication per node , while on regular dht the number of replicas can be controlled . a sample formal specification ( namely : rules ) of the kademlia can be found at @xcite .
note that this specification is given with a tfpl , hence can easily be translated into ontologies .
we can see it has four primitives : ping , store , findnode , and findvalue .
the rules define what to do on each case .
bitcoin is a decentralized p2p network having proofs as its main interpretation , while of course can be written as rules . on @xmath0-chain , every node stores three local ontologies : its routing table , its user s input , and the input from its peers .
those are combined with the ontologies this node subscribes to , from the shared db , together inferring what the client should do now , which can be local storage operations or sending information to peer(s ) .
it is like : i tell you part of what i know , you think about it and tell me the conclusions you want to share with me .
any p2p network is a ruleset defining what to send given what was received .
a p2p network can set its own rules , and many contexts for many rules , and by this on - going implementing new networks over this one , while the user can decide on which contexts they want to participate .
this is @xmath0-chain .
@xmath0-chain has two stages : bootstrapping and maturity .
the network will begin with the following `` bare '' client : it will contain an ontology that implements simple dht , and will run actions by querying the ontology , using cwm , queries like `` what should i do now '' , or more precisely , `` what should i send to each of my peers '' and `` what should i do with my local storage '' . from this point
we can implement everything by ontologies , in a collaborative work , over a decentralized network .
on this bootstrapping stage we plan to insert many readily - made and ontologies , and we plan to gather a round - table of professionals to set this network s global rules together . here
are some possible features to be implemented at the bootstrapping step : * rules to avoid malicious use of the network .
* ability to create separate contexts , where each context has its own rules and they do not interfere . by this , users may create many programs and users can pick which programs they want to use .
like decentralized appstore / google play
. * incentivizing every node for its work . *
implementing bitcoin as a proof of concept . *
distributed large - scale storage , which is essential for the system itself , since it is planned to deal with large amount of data . *
how rules are going to be set from now on .
once the system is thought to have enough strength to go fully public and for everyday use , we can speak of what can be done over it then .
* decentralized source repository ( e.g. decentralized github ) .
* decentralized application repository . *
safely and automatically offer coins - chain will not implement coins from day 1 , but it is something that can and should be built upon the system . ] to a human or machine proving a theorem or writing a software given formal specifications .
* huge db of code fragments ready to be automatically reused by verifying formal requirements or by detecting isomorphisms between ontologies . *
any kind of votes for any purpose , like development team vote for a valid and authentic release , or decentralized democracy . * collaborative social / corporate rule - making , with the ability to find contradictions , to ask `` what is missing in order to obtain x '' . and voting for those rules if wanted .
* fully - customizable , safe , decentralized and private social networks , as well as private clubs ( even with membership fees , entrance test , acceptance rules etc . ) .
* ask human - readable questions about virtually anything . like `` where did aristotles live '' .
cwm is able to answer such questions from ontologies , and specify a proof for its answer . of course ,
other tools are both compatible and much more powerful , e.g. coq .
we have shown how @xmath0-chain is able to generalize any collaborative work , especially peer - to - peer networks .
it provides ultimate information sharing capabilities , with rich ways to query data , infer new information , and act collaboratively .
existing p2p networks can be ported into it , e.g. bitcoin s blockchain , and make them being controlled by additional rules that can be changed on - the - fly with any rule for changing the rules .
it may serve as a universal source of trustable information , as a collaborative source of knowledge , source code , and rules , in a form that is both machine and human readable and processable .
it offers a ground in which sciences can be unified , and , more importantly , people s thoughts can be met and unified cleanly .
formal specification of the kademlia routing table and the kad routing table in maude ,
pita and camacho , 2012 , link : federwin.sip.ucm.es / sic / investigacion / publicaciones / pdfs / kademlia specificationsummary0212.pdf[federwin.sip.ucm.es/sic/investigacion/publicaciones/pdfs/kademlia specificationsummary0212.pdf ] | @xmath0-chain ( pronounced tau - chain ) is a decentralized peer - to - peer network having three unified faces : rules , proofs , and computer programs , allowing a generalization of virtually any centralized or decentralized p2p network , together with many new abilities , as we present on this note . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ `` if law - making is a game , then it is a game in which changing the rules is a move . '' - peter suber presenting nomic @xcite . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ document version : 0.2 feb 6 2015 document version : 0.21 feb 9 2015 document version : 0.22 feb 12 2015 |
the weak gravitational lensing of light from distant sources by the large scale structure in the universe results in the appearance of shear and convergence in images .
the application of ` ray - tracing ' methods , and other programmes , to cosmological @xmath0-body simulations has enabled weak lensing statistics to be recorded for imaginary sources at various redshifts .
these statistics in general show significant differences between different cosmological models , and , of course , differences for sources at different redshifts . from weak lensing magnifications computed from `
ray - tracing ' in @xmath0-body simulations , wambsganss et al .
( 1997 ) were able to estimate the likely magnification biases for sources at redshifts @xmath6 and @xmath7 , and suggested that these biases should be applied to the observed magnitudes of high - redshift type ia supernov .
their work was conducted in a cosmological simulation with density parameter @xmath8 and vacuum energy density parameter @xmath9 , for which the deceleration parameter is @xmath10 .
they found for this cosmology that if the computed median demagnification for @xmath11 were applied to the observed supernov at this redshift it would shift their positions on the hubble diagram indicative of a different cosmology with @xmath12 .
riess et al .
( 1998 ) and perlmutter et al .
( 1999 ) have independently used data sets of type ia supernov extending to a redshift of 0.97 to estimate values of the cosmological parameters from the hubble diagram . in discussion of their results in the light of the work by wambsganss et al .
( 1997 ) , riess et al .
( 1998 ) have stated that the effects on values of the cosmological parameters should be negligible .
perlmutter et al . ( 1999 ) have assumed that the effects of magnification or demagnification average out , and that the most overdense ( or high magnification ) lines of sight should be rare for their set of 42 high - redshift supernov . however , barber et al .
( 2000 ) have now applied a new algorithm developed by couchman , barber and thomas ( 1999 ) for the three - dimensional shear to obtain weak lensing statistics in simulations with cosmological parameters similar to those estimated from the high - redshift supernov data .
their results show that the median demagnification values are considerably less and the dispersions in the probability distributions for the magnification are greater than found by wambsganss et al .
( 1997 ) for redshifts of 1 and 0.5 ( although from different cosmologies ) .
this suggests that the work of wambsganss et al .
( 1997 ) should be reviewed in the light of the new weak lensing data now available from a suitable cosmology . in this paper
i apply the new data from barber et al .
( 2000 ) in an attempt to re - estimate the effects of weak lensing on the cosmological parameters , and find the effects to be far from negligible .
a brief outline of this paper is as follows . in * subsection 1.3 * work by others on adjustments to the cosmological parameters resulting from weak gravitational lensing
is summarised .
* section 2 * summarises the method followed by barber et al .
( 2000 ) for obtaining the weak lensing statistics .
* subsection 2.1 * describes the cosmological @xmath0-body simulations ; * subsection 2.2 * summarises the main features of the algorithm used for the determination of the shear in three - dimensions ; * subsection 2.3 * outlines the method for integrating the shear values throughout the linked simulation volumes , and describes how these values are then used to determine the probability distributions for the magnifications of sources at high - redshift . * section 3 * summarises the relevant weak lensing statistics which are used in the present work . in * section 4
* , i describe the form of the hubble diagram and its dependence on the cosmological parameters , and also summarise the important work by riess et al .
( 1998 ) and perlmutter et al .
( 1999 ) in estimating the cosmological parameters from high - redshift type ia supernov data . in * section 5
* , i discuss how the dispersions in the magnification distributions for different source redshifts may affect the observed distance moduli . for the specific case of the @xmath13 , @xmath4 cosmology
, i then consider by how much the supernov data are displaced on the hubble diagram , and how this translates into effective changes to the perceived values of the cosmological parameters .
* section 6 * summarises the results obtained for the deceleration parameter , and discusses the findings in the light of work by others .
wambsganss et al .
( 1997 ) have used the same ray - tracing method as described by wambsganss , cen and ostriker ( 1998 ) for cosmological simulations with @xmath8 , @xmath9 , and normalisation @xmath14 .
the magnification values above and below which 97% of all of their lines of sight fall were @xmath15 and @xmath16 for source redshifts of @xmath11 , and @xmath17 and @xmath18 for @xmath6 .
the median values for the magnification were 0.983 at @xmath11 , and 0.993 at @xmath19 .
the authors claim that these values would give rise to observed values of @xmath20 for the @xmath11 data , and @xmath21 for the @xmath19 data , rather than the assumed value of @xmath22 .
( the quoted errors arise from the magnification ranges described above in the asymmetrical distributions . ) they also report that the lensing - induced dispersions in their critical density cosmology are three times larger , but this cosmology uses a normalisation of @xmath23 which overproduces the present - day rich cluster abundances . for this cosmology , they report magnifications up to 100 , and correspondingly highly dispersed distributions , very much larger than those of barber et al .
( 2000 ) for @xmath24 .
the high magnifications almost certainly derive from the low value of their ( fixed ) softening scale .
fluke , webster and mortlock ( 1999 ) and fluke , webster and mortlock ( 2000 ) have used a ` ray - bundle ' method in which a discrete bundle of light rays is traced .
the method allows a direct comparison between the shape and size of the bundle at the observer and at the source plane , so that the magnification , ellipticity and rotation can be determined straightforwardly .
the cosmological models investigated were similar to those of barber et al .
however , the particles were considered as point masses , so that very high magnification values could be achieved in principle ; however , to alleviate this situation , the authors do not include bundles which pass within @xmath25 of the einstein radius of any particle . because of their use of point particles , they use the empty beam approximation which gives rise to magnification probability distributions with minimum magnifications of 1 ( and therefore mean and median values greater than , or equal to , one ) , high magnification tails , and broad dispersions in magnification . for @xmath11 ,
the distribution in the magnifications for the scdm cosmology ( @xmath26 @xmath27 ) is clearly broader than that for the low density ( lcdm ) cosmology ( @xmath13 , @xmath4 ) , which in turn is broader than the open ( ocdm ) cosmology ( @xmath28 @xmath27 ) .
this is also true of barber et al.s ( 2000 ) data , although barber et al .
( 2000 ) find a completely different order for the cosmologies at higher source redshifts . because of the high magnification tails in fluke , webster and mortlock s ( 2000 ) distributions
, they discard those high magnification lines of sight which occur with low probability before defining the @xmath29 and @xmath30 values .
these magnification probability distributions are then used by fluke and webster ( 2000 ) to examine the effects of the weak lensing dispersion on measurements of @xmath31 in both the empty beam ( more appropriate for point particles ) and full beam limits .
they consider whether the resulting shifts in the hubble diagram can be fitted with various cosmological models . with input data from the @xmath7 magnification distribution for the flat lcdm simulations , they find that @xmath32 from the empty beam distribution . to convert their empty beam magnification values to filled beam values , fluke and webster ( 2000 ) use a simple scaling relationship .
however , since the filled beam approach is inappropriate to their method , the resulting values should be treated with caution .
they find the revised value @xmath33 ; i.e. , a much larger departure than for the empty beam approximation , but a similar dispersion . again
the dispersion can be partly explained in terms of their use of point particles , although they do attempt to modify their distributions to account for the high magnification tails .
it should be mentioned that wambsganss et al .
( 1997 ) have assumed that the observed supernov have magnifications of unity , and then have assessed departures from the best fit cosmology , rather than assuming that the best fit cosmology comes from the most likely demagnification at the peak of the magnification distributions , which is the approach i take in this paper .
although fluke and webster ( 2000 ) use the median magnification values , the values are always close to unity and greater than , or equal to , unity for the empty beam case .
the cosmological @xmath0-body simulations used in the study by barber et al .
( 2000 ) were provided by the hydra consortium and produced using the ` hydra ' @xmath0-body hydrodynamics code ( couchman , thomas and pearce , 1995 ) .
results are reported on simulations from four different cosmologies , referred to as the scdm , tcdm , ocdm and lcdm cosmologies .
each of the simulations uses a cold dark matter - like spectrum , and the parameters used in the generation and specification of these simulations are listed in table [ cosmo ] .
@xmath34 and @xmath35 are the present - day values of the density parameter and the vacuum energy density parameter respectively .
the power spectrum shape parameter , @xmath36 , is set to 0.5 in the scdm cosmology , but the empirical determination ( peacock and dodds , 1994 ) of 0.25 for cluster scales has been used in the other cosmologies . in each case , the normalisation , @xmath37 , on scales of @xmath38mpc has been set to reproduce the number density of clusters ( viana and liddle , 1996 ) ; @xmath39 is the hubble parameter expressed in units of 100 km s@xmath40mpc@xmath40 .
.parameters used in the generation of the four different cosmological simulations . [ cols="^,^,^,^,^,^,^ " , ] @xmath41 riess et al .
( 1998 ) have recognised that weak lensing of high - redshift type ia supernov can alter the observed magnitudes , and quote the findings of wambsganns et al .
( 1997 ) i.e. , that the light will be on average demagnified by 1.7% at @xmath7 in a universe with a non - negligible cosmological constant .
they state that the size of this effect is negligible .
perlmutter et al .
( 1999 ) assume that the effects of magnification or demagnification will average out , and that the most over - dense lines of sight should be rare for their set of 42 high - redshift supernov .
however , they note that the average ( de)amplification bias from integration of the probability distributions is less than 1% for redshifts of @xmath42 .
my work has shown that , in the lcdm cosmology , a source at @xmath7 will have a median deamplification of 3.4% ( @xmath43 ) , and 1.3% at @xmath6 .
whilst these demagnification factors are still small , they can result in repositioning of the supernov data on the hubble diagram with significantly revised values of the cosmological parameters .
in addition , there is a significant probability of observing highly magnified supernov ; 97% of all lines of sight display a range of magnifications up to 1.191 at @xmath7 , and this range is equivalent to a dispersion in the magnitudes of 0.252 .
the standard deviation in the asymmetrical distributions at @xmath11 is 0.078 magnitudes .
this is to be compared with the accepted intrinsic dispersion of 0.17 magnitudes for type ia supernov reported by hamuy et al .
( 1996 ) in a set of low - redshift supernov . for sources at a redshift of 2 , however , barber et al.s ( 2000 ) data predict a standard deviation of 0.185 magnitudes from weak lensing , in excess of the intrinsic dispersion
. clearly , the weak lensing dispersion will become increasingly important as supernov at greater redshifts are discovered . in the work by wambsganss et al .
( 1997 ) in the cosmology with @xmath8 and @xmath9 they found @xmath15 and @xmath16 for source redshifts of @xmath11 , and @xmath17 and @xmath18 for @xmath6 .
their median magnification values were 0.983 for @xmath11 and 0.993 for @xmath19 .
these may be compared directly with the values shown in table [ mu4c ] for the magnifications in the @xmath44 @xmath4 cosmology i have investigated , i.e. , @xmath45 and @xmath46 for source redshifts of @xmath11 , and @xmath47 and @xmath48 for @xmath6 .
the median values are 0.966 for @xmath11 and 0.987 for @xmath49 .
since the most likely effect of weak lensing is a slight demagnification of the source , it is probable that the originating data used by riess et al .
( 1998 ) and perlmutter et al .
( 1999 ) may suffer from such a demagnification , and therefore should be re - positioned on the hubble diagram at brighter positions ( i.e. , lower distance moduli ) .
this would then point to a slightly different cosmology with a different deceleration parameter .
i plot , in figure [ modulus / omega ] , the distance modulus values at @xmath7 , for various values of @xmath34 , centred around @xmath13 , and keeping @xmath50 .
the plot has been generated by solving equation [ modulus ] numerically .
@xmath51 with no magnification , @xmath52 @xmath4 , @xmath53 , and @xmath11 , the distance modulus is 44.2626 .
now assume that this cosmology is the best fit to the observed data .
then , by making a redshift 1 supernova brighter than observed ( assuming it was observed at the peak of the magnification distribution , with @xmath43 ) the distance modulus would be reduced to 44.2250 , equivalent to @xmath54 , and equivalent to changing @xmath31 from @xmath55 to @xmath56 .
for the rms deviations in the magnification , the distance modulus changes by @xmath57 , equivalent to revised @xmath31 values of @xmath58 and @xmath59 .
for the @xmath30 and @xmath29 values at @xmath7 , @xmath31 changes to @xmath60 and @xmath61 .
the conclusion , on the evidence of these data , is that true underlying cosmologies with @xmath62 may be interpreted as having @xmath2 , from the use of perfect standard candles ( without intrinsic dispersion ) , arising purely from the effects of weak lensing .
this adjustment to the value of @xmath31 is almost an order of magnitude larger than that found by wambsganss et al .
( 1997 ) based on a cosmology with @xmath63 @xmath9 , and the dispersion in the values is slightly larger than their findings .
i have summarised the results for the magnification distributions for the different cosmologies for different source redshifts as obtained by barber et al . (
2000 ) in table [ mu4c ] . at high redshift
, the lcdm cosmology produces the highest magnifications , the broadest distribution curves , and the lowest peak values . for sources at @xmath64 in the lcdm cosmology , 97% of all lines of sight have magnification values up to 1.850 .
( the maximum magnifications , not quoted here , depended on the choice of the minimum softening in the code , although the overall distributions were very insensitive to the softening . ) the rms fluctuations in the magnification ( about the mean ) were as much as 0.191 in this cosmology , for sources at @xmath64 . even for sources at @xmath6
there is a measurable range of magnifications in all the cosmologies .
the immediate implication is the likely existence of a bias in observed magnitudes of distant objects , and a likely dispersion for standard candles , for example , type ia supernov at high redshift . in particular , the weak lensing dispersion from 97% of the lines of sight expected in the peak magnitudes for type ia supernov at redshifts of 1 may be as much as 0.252 magnitudes .
the standard deviations in the distributions are 0.078 magnitudes for @xmath11 and 0.185 magnitudes for @xmath65 .
these values are to be compared with the accepted dispersion of 0.17 magnitudes , so that we should expect to see an increasing dispersion for the peak magnitudes of supernov as they are discovered at higher redshifts .
i have made use of the magnification statistics from the lcdm cosmology , to reanalyse the results for the cosmological parameters determined from the high - redshift type ia supernov data .
( both riess et al . , 1998 , and perlmuter et al . , 1999 ,
point to cosmologies with parameters close to those of the lcdm model , i.e. , @xmath13 , @xmath66 . )
the dispersions in the magnification are monotonically increasing with redshift , and have rms fluctuations about the mean of 0.045 for @xmath67 , and 0.016 for @xmath68 .
these have been translated into variations in distance modulus , which , in turn , suggest a different cosmological model for the data . on the assumption that @xmath3 , @xmath4 and @xmath2 is the cosmology which fits the observed data best ,
my results indicate that the true underlying cosmology could have @xmath69 based on the peak magnification values , with a spread of @xmath70 based on the computed magnification dispersion .
this would represent a significant adjustment to the cosmological parameters .
wambsganss et al .
( 1997 ) found a median demagnification value much closer to unity ( 1.7% below for @xmath11 ) in their cosmology with @xmath63 @xmath9 than barber et al .
( 2000 ) have found ( 3.4% below for @xmath11 ) in the @xmath13 , @xmath4 cosmology .
consequently , my adjustment to @xmath31 is considerably greater .
however , the magnification dispersions give rise to similar dispersions in @xmath31 .
wambsganss et al.s ( 1997 ) calculated adjustment to @xmath31 is approximately an order of magnitude smaller than the departure i would suggest .
fluke , webster and mortlock s ( 2000 ) estimate for the revised value of the deceleration parameter ( @xmath71 for the @xmath72 @xmath4 cosmology ) in the empty beam approximation is only half the value of the departure i suggest .
the value of their dispersion can not be easily compared with mine , because of their use of the empty beam approximation , which gives rise to a different form for the distribution curve with values of magnification which are always positive . furthermore , their method to convert the empty beam magnification values ( more appropriate to their method ) to filled beam values was not rigorous , so that the resulting values should be treated with caution . both wambsganss et al .
( 1997 ) and fluke , webster and mortlock ( 2000 ) have assumed that the observed supernov have magnifications of unity , so that the median demagnifications when using the filled beam approach are assumed to reposition the supernov at dimmer magnitudes .
fluke , webster and mortlock ( 2000 ) have to reposition them at brighter magnitudes when using the empty beam approximation .
i have adopted the opposite viewpoint .
i have assumed that the supernov are observed at their median demagnification values , and therefore have to be repositioned at brighter magnitudes ( i.e. , reduced distance moduli ) to obtain the correct cosmological parameters .
consequently , the direction of my adjustment to @xmath31 is opposite to those of the above authors for the filled beam approximation .
i am indebted to the starlink minor node at the university of sussex for the preparation of this paper , and to the university of sussex for sponsorship .
i have benefited from useful exchanges with dr .
peter thomas of the university of sussex and dr .
chris fluke of swinburne university .
0.41 cm perlmutter , s. , gabi s. , goldhaber g. , goobar a. , groom d. e. , hook i. m. , kim a. g. , kim m. y. , lee j. c. , pain r. , pennypacker c. r. , small i. a. , ellis r. s. , mcmahon r. g. , boyle b. j. , bunclark p. s. , carter d. , irwin m. j. , glazebrook k. , newberg h. j. m. , filippenko a. v. , matheson t. , dopita m. and couch w. j. , 1997 , ap .
j.,483 , 565 .
perlmutter s. , aldering g. , goldhaber g. , knop r. a. , nugent p. , castro p. g. , deustua s. , fabbro s. , goobar a. , groom d. e. , hook i. m. , kim a. g. , kim m. y. , lee j. c. , nunes n. j. , pain r. , pennypacker c. r. , quimby r. , lidman c. , ellis r. s. , irwin m. , mcmahon r. g. , ruiz - lapuente p. , walton n. , schaefer b. , boyle b. j. , filippenko a. v. , matheson t. , fruchter a. s. , panagia n. , newberg h. j. m. , couch w. j. , and project t. s. c. , 1999 , ap . j. , 517 , 565 .
riess a. g. , filippenko a. v. , challis p. , clocchiatti a. , diercks a. , garnavich p. m. , gilliland r. l. , hogan c. j. , jha s. , kirshner r. p. , leibundgut b. , phillips m. m. , reiss d. , schmidt b. p. , schommer r. a. , smith r. c. , spyromilio j. , stubbs c. , suntzeff n. b. and tonry j. , 1998 , a. j. , 116 , 1009 . | significant adjustments to the values of the cosmological parameters estimated from high - redshift type ia supernov data are reported , almost an order of magnitude greater than previously found .
they arise from the effects of weak gravitational lensing on observations of high - redshift sources .
the lensing statistics used have been obtained from computations of the three - dimensional shear in a range of cosmological @xmath0-body simulations , from which it is estimated that cosmologies with an underlying deceleration parameter @xmath1 may be interpreted as having @xmath2 ( appropriate to the currently popular cosmology with density parameter @xmath3 and vacuum energy density parameter @xmath4 ) .
in addition , the standard deviation expected from weak lensing for the peak magnitudes of type ia supernov at redshifts of 1 is expected to be approximately 0.078 magnitudes , and 0.185 magnitudes at redshift 2 .
this latter value is greater than the accepted intrinsic dispersion of 0.17 magnitudes .
consequently , the effects of weak lensing in observations of high - redshift sources must be taken properly into account .
3mp@xmath5 m # 1*#1 * # 1#2#3#4#5#1 , # 2 , # 3 , # 4 , # 5 # 1#2#3#1 , # 2 , # 3 , in press # 1#2#3#4#5#6#1 , # 2 , in # 3 , eds , # 4 .
# 5 , p. # 6
# 1#2#1 , # 2 , preprint = cmti7 # 1 # 1 = wncyr10 at 12pt = wncyr7 at 8.5pt = wncyr5 at 6pt = = = # 1#22.1pt-1.2pt/-1.2pt 2.1pt # 1*#1 * galaxies : clusters : general cosmology : miscellaneous cosmology : gravitational lensing large - scale structure of universe cosmology : observations supernovae : general |
until the detection of an azimuthal color gradient across one of the arms of the sac galaxy m 99 ( * ? ? ?
* gg96 hereafter ) , only sparse evidence of star formation triggered by spiral density waves had been found in stellar counts in the milky way@xcite and m 31@xcite .
more recently , applying the same method defined by gg96 and described below , ( * ? ? ?
* mg09 hereafter ) examined a sample of 13 spiral galaxies of types a and ab , and found color gradients consistent with theoretical expectations in 10 of their objects .
although they did not compute stellar orbits , both gg96 and mg09 implicitly assumed that all stars , including those recently born near spiral arms , move in circular orbits .
mg09 did investigate the effects of variable circular speeds and variable densities on observed color gradients and found them to be negligible .
however , complicated non - circular motions have been reported in studies about the migration of young stars following star formation triggered by spiral shocks @xcite .
@xcite had also noticed that the ballistic orbits calculated from galactic hii region complexes follow non - circular trajectories that initially move along ( and not across ) the spiral arms . in the present work ,
we examine how the presence of non - circular motions would modify the pattern speeds derived from color gradients under the assumption of circular orbits .
we will explore two different approaches .
the first one ( see [ appr1 ] ) involves the semi - anaylitical solutions obtained for spiral shocks @xcite ; we assume that newly born stars preserve the orbital motion of the shocked gas where they form .
the second approach ( see [ appr2 ] ) is based on data from mhd simulations ; we follow the gas flow vectors near spiral arms .
we always assume that stars are triggered all along the studied regions of spiral arms in the respective model .
other methods , not discussed here , may involve orbit calculations for young stellar groups .
the photometric technique employed by gg96 and mg09 uses three optical bands , @xmath1 , @xmath2 and @xmath3 , plus the near infrared @xmath4 band ( see table [ tbl - filters ] ) . with these filters ,
the reddening insensitive and star - formation sensitive photometric index @xmath5 is obtained : @xmath6 @xmath7 in star - forming regions , the @xmath8 index has higher values , because the _ g _ and _ j _ bands in the numerator of equation [ eqqlog ] trace the light from red and blue supergiants , respectively .
when the star - formation activity is poor , the value of the @xmath8 goes down .
@xmath1 _ & 5000 & 830 + _ @xmath2 _ & 6800 & 1330 + _ @xmath3 _ & 7800 & 1420 + _ @xmath4 _ & 1.25 & 0.29 + for each studied galaxy and at each wavelength , the spiral arms are first unwrapped @xcite , by plotting them in a @xmath9 versus @xmath10 map ( see figures [ spiral ] , [ unwra ] ) .
all the spiral arms in the disk galaxy are inspected , in search of color gradients .
if a candidate region is found , the arm is straightened , by adding a different phase - shift to @xmath10 at each value of @xmath9 , until the arm appears as a horizontal line ( see figure [ strai ] ) .
this procedure allows us to add the data from different @xmath9 , and thus to increase the signal - to - noise ratio of the light profile as a function of @xmath10 .
1-d plots of @xmath5 vs. @xmath10 are obtained from the unwrapped and straightened galaxy images .
these are then compared to 1-d plots of @xmath11 , that trace the dust lane location , and to 1-d plots of the @xmath12-band ( or @xmath13 or @xmath14 , depending on what we have available ) data , that trace the density wave ( under the assumption that the near - ir emission of red supergiants can be neglected ) .
the regions that seem to match theoretical expectations according to these plots are selected and compared to stellar population synthesis models , that provide @xmath5 as a function of stellar age , @xmath15 .
if one assumes that stars form in the site of the shock , and that they age as they move away from this location , then distance from the dust lane ( at constant radius ) parametrizes stellar age .
in fact , stretching the model @xmath16 to fit the data ( where @xmath5 is a function of angular or linear distance ) fixes the ratio between the distance from the shock , @xmath17 , and the age of the stellar population .
if , in addition , the rotational velocity is known , it is possible to find the angular velocity of the spiral pattern , @xmath18 , as follows : @xmath19 here , @xmath20 is the mean radius of the region where the gradient has been detected , measured from the center of the galaxy ; @xmath21 is the rotational velocity of the disk at such radius ( obtained from the literature ) ; @xmath17 is distance from the shock , and @xmath15 is the stellar age obtained from fitting the population synthesis model to the observations . with the pattern speed , we obtain the location of major resonances : the outer lindblad resonance ( olr ) , corotation , and the 4/1 resonance .
this resonance positions are compared with the observed spiral end points . as mentioned before
, this procedure implicitly assumes circular motion for the stars involved , even though no stellar orbits have been computed .
shock solutions obtained from semi - analytical approaches have shown that gas streamlines are not circular .
@xcite , for example , found that streamlines appear as sharp - pointed ovals ( see figure [ sharp ] ) .
although such solutions were found for the tightly wound approximation , they provide a good reference for the study of real galaxies ( see also * ? ? ?
, we obtain gas streamlines under the assumption that young stars follow the motion of the molecular gas , and by adopting a procedure similar to that of @xcite .. the reference system is the rotating frame of the spiral pattern . ] our adopted model has a pattern speed of 13 km s@xmath22 kpc@xmath22 , and a flat rotation curve with @xmath23 km s@xmath22 . from @xmath24 kpc to @xmath25 kpc
, streamlines were calculated in steps of 0.05 kpc .
since the stellar models give us @xmath5 as a function of @xmath26 , we can assign a stellar age to each point in the streamlines ; in order to do so , we need to assume that the onset of star formation occurs at a certain orbital time @xmath27 , where @xmath28 corresponds to the shock position .
@xcite use the term
timescale for star formation " as the duration of the burst , that is estimated to be @xmath29 years .
they do not make any assumptions about the time it takes for star formation to begin after the shock .
a preliminary inspection of their findings shows that the shock ( as traced by the dust lanes ) and the onset of star formation ( marked by @xmath30 in the stellar models ) are not always at the same location . in real galaxies , whether there is a delay between the shock and the star formation onset is not well understood yet .
in fact , the time needed for a diffuse cloud of neutral gas to first become a dense cloud , then a molecular cloud , and finally a self - gravitating cloud is @xmath31 years .
once this cloud is formed , the onset of star formation may be very fast @xcite .
@xcite , based on an observational study between the peaks of molecular ( @xmath32 ) and young stellar ( @xmath33 ) arms , report a delay for the onset of star formation of @xmath34 myr in 5 galaxies ( out of 13 ) ; @xcite estimate a timescale for star formation in the range 1 - 4 myr , from a study of angular offsets between hi and 24@xmath35 emissivity peaks in a sample of 14 disk galaxies .
in order to test for the effects of a star formation onset delay on azimuthal color gradients , we try three cases . the first one has a delay ( i.e. , the time it takes for the gas to move from the shock to the location where star formation begins ) @xmath36 yr ; the other two have @xmath37 yr and @xmath38 yr , respectively .
the gas simulations were performed with a version of the zeus code @xcite , that is an eulerian , time - explicit , finite - difference code for ideal mhd simulations .
we employ a 2d grid with 500 @xmath39 500 points in polar coordinates .
the radial extent goes from @xmath40 3 to 30 kpc , and the azimuthal one from @xmath41 to @xmath42 radians .
we assume that the half disk simulation data have a @xmath43 `` mirror '' symmetry .
the gaseous disk follows an isothermal equation of state with a temperature of 10900@xmath44 k. at the beginning of the simulation , the magnetic field has a toroidal geometry and a value of @xmath45 g at @xmath46kpc from the galactic center , although it rapidly evolves away from this setup .
no self - gravity is included .
the background gravitational potential has two components : one axisymmetric , that consists of a bulge , a disk , and a halo , while the other is the non - axisymmetric spiral arm perturbation .
the adopted bulge and disk are described in @xcite .
the dark matter halo is nfw type @xcite , with density : @xmath47 where @xmath48=@xmath49 @xmath50 and @xmath51=15.133 kpc .
the two - arm spiral potential has a pitch angle of @xmath52 , and is self - consistent in the stellar orbits sense @xcite .
the simulation is performed in this stellar arms reference frame , that rotates with velocity @xmath53 kms@xmath22kpc@xmath22 .
we allow the simulation to evolve during @xmath54yr ( see figure [ gilgosim ] ) . by this time ,
the gas develops spiral arms as a response to the imposed perturbation , initial transients subside , and changes in the simulation are only observable at very long time scales .
therefore , we assume that density and velocity distributions at this time represent a close approximation to a steady state solution .
as shown in figure [ gilgosim ] , the gas in the simulation has responded to the spiral perturbation with a four arm pattern @xcite .
a hypothetical circular gas orbit would encounter four shocks before completing its transit around the disk . in this case
, the corresponding velocity field would be very different from the case where only two shocks are considered .
for the present investigation we focus on the main shock ( i.e. , the one closer to the potential minimum of the spiral perturbation ) , although spiral - arm triggering of star formation may also take place in the secondary shocks . in order to obtain the orbit a parcel of gas would follow ,
we locate the main shock position by searching for the gas density maxima closest to the ( stellar ) spiral arms .
these are taken as the start point for the integration of the gaseous orbits .
the orbits are then calculated from the velocity data ( in the spiral pattern reference frame ) , using a runge - kutta method , every @xmath55kpc , and @xmath56s .
a sample of orbits are shown in figure [ gorbits ] .
other gas orbits with mhd simulations using the full time evolution of the gas can be seen in @xcite , who find a close resemblance between gas orbits and the central family of ballistic orbits in stellar dynamics .
from the first approach , involving semi - analytical solutions , we obtain @xmath5 index profiles for the non - circular and circular cases , respectively , calculated in a narrow annulus with mean radius @xmath57 kpc ; we show them in figure [ diff_q ] . to produce the @xmath5 profiles , we assume a background population of old stars with an age of @xmath58 yr ( i.e. , the density wave ) , plus a young burst of star formation lasting @xmath59 yr .
both the old and the young populations have a salpeter imf , with @xmath60 and @xmath61 . for the present case , we adopt a constant fraction of young stars of 2% by mass ( see equation 9 in mg09 ) .
we then use the stellar population synthesis models by @xcite to find the changing @xmath1 , @xmath2 , @xmath3 , and @xmath4 emission of the population mixture as it evolves . in the circular case ,
the stellar complexes that give origin to the @xmath5 profile were all born at the same ( @xmath62 ) position . in the non - circular case , due to the trajectories of the gas streamlines , the stars that give rise to the @xmath5 profiles
were not all formed at the same location .
the age of the stellar complexes that contribute to the @xmath5 profile as a function of distance from the shock position is plotted in figure [ diff_t ] .
as expected , young objects tend to spend more time concentrated toward the spiral shock ( @xmath63 kpc ) , with the consequence that the peak of the @xmath5 profile ( _ dashed line _ in figure [ diff_q ] ) occurs closer to the shock .
another aspect to be noticed in figure [ diff_q ] is that there is no downstream fall " of the gradients ( i.e. , @xmath5 does not fall below , in this case , @xmath64 , contrariwise to what is sometimes observed with the data ) .
@xcite had hypothesized that the fall of the observed @xmath5 profiles below the model values ( assuming pure circular orbits ) might be caused by stellar non - circular motions in the data .
however , results may differ if ballistic trajectories with postshock velocities are considered ( e.g. * ? ? ?
* ; * ? ? ?
* ) . from the @xmath5 profiles , it is possible to derive pattern speeds by comparing stellar population synthesis models ( @xmath5 vs. @xmath15 ) to the data " ( @xmath5 vs. @xmath17 ) , assuming implicitly that stars move in circular orbits , as was done by gg96 and mg09 , and was sketched above in section [ themethod ] . in the case of the semi - analytical approach ,
data " @xmath5 profiles were obtained from narrow annular regions , @xmath65 kpc wide .
given that we know the input pattern speed , we can search for systematic effects in the determination of @xmath18 .
figure [ delayt ] shows the input pattern speed ( _ long - dashed line _ ) , and @xmath0 values that are obtained from the color gradients for three different presuppositions about the delay of star formation onset after the shock ( 0 , 10 , and 20 myr ) . for reasons explained below in
[ lnrcoll ] , rather than stretching the stellar population synthesis model to the data " , we just compare the positions of the maxima of the model @xmath5 , on the one hand , and of the shock location , on the other . there is a significant systematic effect , whereby the derived @xmath0 is always larger than the input pattern speed ; moreover , the effect decreases with galactocentric radius .
also , the difference between the input and the output pattern speeds decreases for larger star formation onset delays . in the case of the mhd simulations
, we again analyze the pattern speeds derived when assuming a circular motion dynamic model . in order to obtain @xmath0
, we compare the peak value of @xmath5 in the stellar population model to the density maximum of the gas in the simulations . in figure
[ apcm ] , we show the @xmath18 values obtained for @xmath66 of 0 , 10 , and 20 myr .
one of the caveats of the gg96 method , that was not discussed in mg09 , comes from image processing .
each pixel in the optical and infrared images actually contains information about many young stellar orbits that all fall within the same spatial region .
the process of unwrapping " the spiral arms then averages pixels in the @xmath10 and @xmath67 directions ( the straightening " of the arms is just a shift in the @xmath10 direction and does not involve pixel averaging ) . in order to increase the signal - to - noise ratio of the @xmath5 profiles , an additional averaging in the @xmath67 direction
is done in selected regions .
this last step averages together orbits with different angular speeds , and hence stars of slightly different ages at a fixed distance from the shock .
figure [ cavfig ] shows the behavior of @xmath5 profiles from an unwrapped and straightened synthetic image ( see below ) , under the assumption of pure circular motion ; the different profiles are taken from various mean radii , after averaging in @xmath67 .
the shift of the peak toward @xmath63 is expected as we approach the corotation radius . however , due to the pixel averaging , a drop in the maximum value of @xmath5 is also obtained . if , in order to match the observed or synthetic profile peak , a constant downward shift is applied to the stellar population synthesis model , the `` wings '' of the profiles will not match , since those of the stellar model will now lie within the `` wings '' of the data .
this mimics wider gradients , and in order to fit the observed or synthetic profile we will need to overstretch the model .
consequently , the pattern speed obtained via equation [ eqomega ] will be lower , because the stellar age ( @xmath15 ) at a fixed distance @xmath17 from the shock will be underestimated . the effect amounts to @xmath68 1 to 2 km s@xmath22 kpc@xmath22 for moderate pixel averaging , but can be as large as 10 km s@xmath22 kpc@xmath22 when the averaged region in @xmath67 is very extended .
one way out of this problem is to compare only the distances between the onset of star formation and the @xmath5 peak , in the model and the data , instead of the whole @xmath5 profiles , since image processing affects less the peak positions than the profile shapes .
we have used this approach with our semi - analytical calculations and mhd simulations ; unfortunately , in the case of real data , the position of the star formation onset is generally unknown ( see also [ appr1_sub ] ) . to better estimate the effects of averaging , with the @xmath69 , @xmath70 and @xmath27 ( time elapsed since the shock
, @xmath27=0 corresponds to the shock location ) data of the streamlines obtained from the semi - analytical solutions , we constructed images of a synthetic galaxy " as it would be observed in the @xmath1 , @xmath2 , @xmath3 and @xmath4 bands with the aid of stellar population synthesis models @xcite ( the fraction of young stars constitute is @xmath71 by mass ) .
the model face - on galaxy is at a distance of 35 mpc and the synthetic images have @xmath72 pixels .
whenever various streamlines were located at the same pixel , their elapsed times since the shock were averaged .
next , the synthetic images were unwrapped .
each hyperpixel of the unwrapped images is @xmath73 long in the @xmath10 axis ; the @xmath67 axis has 250 hyperpixels in total , from the center of the galaxy to @xmath25 kpc .
then , the synthetic unwrapped images were straightened according to the shock s pitch angle ( @xmath74 ) , that is smaller than the potential minimum s pitch angle ( @xmath75 ; see appendix [ appsemianalyt ] ) . finally , @xmath5 profiles were obtained from these images " in sections with a width of 5 hypercolumns in @xmath67 .
pattern speeds were measured by comparing the observed " @xmath5 ( vs. @xmath17 ) from the simulated images , at different radii , with @xmath5 ( vs. @xmath15 ) from stellar population synthesis models .. models with higher upper limits would peak at younger ages , closer to the shock ; see mg09 for models with an imf upper mass limit of 100 @xmath76 . ] in figure [ myomegas ] , we illustrate the effects of averaging .
the dotted line is the input pattern speed .
if the synthetic galaxy has circular orbits and the galaxy is analyzed assuming circular orbits , then one obtains the @xmath0 values depicted by squares .
the solid black squares are obtained with the gg96 stretching " method , whereas the comparison between model and data @xmath5 maxima positions relative to the shock yields the open symbols .
clearly , the solid black squares are biased while the open ones are not . on the other hand , if the synthetic galaxy has non - circular orbits but is analyzed under the assumption of circular orbits , then one obtains the @xmath18 values shown as triangles .
the solid black triangles are the values obtained with the gg96 method ; the empty triangles are the pattern speeds found when comparing only the positions , relative to the shock location , of model and observed " @xmath5 peaks ; and the solid line represents the biased values measured from the semi - analytical solutions , without averaging , if existing non - circular motions are ignored .
the reason for this bias towards higher measured pattern speeds is that real gas streamlines in a steady rotating spiral shock turn somewhat along the arm after passing through the shock .
consequently , stars close to the shock are slightly older than would be expected in a circular model with the same @xmath0 .
the observer using a circular model would infer that the gas flow into the arms is smaller than is actually the case , i.e. , that the difference between the stellar orbital velocity and the pattern speed is smaller than in reality ; inward of corotation , this means that the observer would overestimate the pattern speed . also inside corotation ,
the effect decreases with galactocentric radius , as the shock strength diminishes .
the open triangles product of the peak comparison are much less sensitive to the competing systematics introduced by orbit averaging .
however , ironically , the original stretching " method ( solid black triangles ) seems much less biased towards higher values if non - circular motions are neglected , although the dependence of the bias on galactocentric radius is still noticeable .
this is a fortunate turn of events , because ( 1 ) in real images it is hard to pinpoint the location of the star formation onset , and ( 2 ) the radial dependence of the bias allows us to detect the presence of non - circular motions and to confirm the link between star formation and disk dynamics .
from theory ( e.g. , * ? ? ?
* ) , it is well known that the spiral shock strength diminishes as one approaches corotation .
the shock weakening implies fewer radial movements and more circular trajectories .
figures [ delayt ] ( corotation radius , @xmath77 , located at @xmath78 kpc ) and [ apcm ] ( @xmath79 kpc ) support this conclusion ; they also allow us to predict that , in general , @xmath0 values derived from color gradients observed inside corotation in real data will be too high if a circular dynamic model is adopted . 6 and 8.5 kpc ) in the mhd simulation , the measured @xmath18 will be actually slightly lower ( @xmath68 1 km s@xmath22 kpc@xmath22 ) than the input pattern speed of 20 km s@xmath22 kpc@xmath22 . ] for color gradients observed beyond corotation , we expect again to overestimate the pattern speed , but the difference between the actual and the measured @xmath18 will now grow with radius . in this case , the radial velocities of the streamlines after the shock have an outward , rather than an inward , component ( see leading case in * ? ? ?
if , however , spiral shocks are not as strong as inside corotation , inverse post - corotation gradients will yield unbiased pattern speed measurements .
how would these results affect the @xcite conclusions about the end points of spiral patterns ?
their figure 4 displays the ratio , @xmath80 , of the resonance radii ( calculated from their pattern speed measurements and rotation curves ) to the observed spiral end points in their deprojected , near - infrared galaxy images
. if their @xmath18 measurements ( most of them inside corotation ) are systematically too high , then the resonance radii , and their ratio to the spiral end points would be systematically smaller than the true values . in order to better quantify the impact of non - circular motions on the mg09 results , we define the expression @xmath81 . here
, @xmath82 is the pattern speed obtained from the data , by comparing the observed color gradient candidates with the stellar models , under the assumption that stars move in purely circular orbits ; @xmath83 is the pattern speed that the spiral should have if it ends at certain resonance . for the 4:1 resonance , for example
, we have : @xmath84 where @xmath85 , is the circular orbital velocity ( obtained from the literature , see mg09 ) . and
if the spiral ends at the olr : @xmath86 to calculate @xmath87 for the 23 regions analyzed in mg09 , we take both @xmath82 and @xmath88 from their table 4 . in figures [ diffom_4_1 ] and [ diffom_olr ] , we plot @xmath87 vs. @xmath89 , where @xmath90 is the mean orbital radius of the studied region .
figure [ diffom_4_1 ] corresponds to the case where the end point of the spiral pattern is fixed at the 4:1 resonance .
@xcite state that strong spirals ( hubble types sb or sc ) may be truncated at this resonance .
the @xmath87 values show the trend with radius expected from figures [ delayt ] and [ apcm ]
. this would imply that we are actually detecting non - circular motions from the photometric data ! however , three discrepancies with respect to theoretical expectations are also present .
the first one is related to the systematic effects discussed in appendix [ lnrcoll ] . in spite of the tendency to actually measure values around the true pattern speed with the method used by gg96 and mg09 ( solid black triangles in figure [ myomegas ] ) ,
there is a lack of points below the @xmath91 line .
the second difference is the gap around @xmath92 , where a point with huge vertical error bars is located .
if spirals are really truncated at the 4:1 resonance , and hence inside or near corotation , one would expect a continuous distribution of points between the ilr and the spiral end point ( i.e. , @xmath93 ) .
the third inconsistency concerns the actual values attained by @xmath87 at low @xmath89 . from the solid black triangles in figure [ myomegas ] , the maximum expected value of @xmath87 is @xmath94 .
the high @xmath87 values would imply the presence of extremely strong shocks near the spiral arms .
figure [ diffom_olr ] , in contrast , shows the behavior of @xmath87 for the case when the spiral end points are fixed at the olr .
if spirals stop near this resonance , the expected @xmath89 value for corotation is @xmath95 , and the gap in the data distribution is naturally explained .
we also observe now that some points are located below the @xmath91 line , as expected from figure [ myomegas ] , owing to the systematic error introduced by the stretching " gg96 method and discussed in [ lnrcoll ] .
the three last points , marked with a long - dashed ellipse , represent the `` inverse color gradients '' ( i.e. , located beyond corotation ) examined in mg09 ( these points were treated as being inside corotation in figure [ diffom_4_1 ] ) .
the points surrounded by a short - dashed ellipse belong to the galaxy ngc 578 .
if ngc 578 actually ends before or at corotation ( as argued by mg09 ) , its position in figure [ diffom_4_1 ] is in concordance with theoretical expectations ( i.e. , close to the @xmath91 line ) . on the other hand
, ngc 578 may be the only galaxy in the mg09 sample with a spiral pattern that stops at the 4:1 resonance , as predicted by @xcite .
under the assumption that the orbits of young stars preserve the velocity components of the parent molecular clouds where they form , we have analyzed the effects that non - circular motions would have on azimuthal color gradients .
semi - analytical calculations and mhd simulations show that the spiral pattern speeds derived from the comparison between color gradients and stellar population synthesis models , assuming purely circular motions , would have values systematically higher than the real ones for regions within corotation . also inside corotation
, the effect would decrease with galactocentric radius .
non - circular motions , however , would not prevent the detection of legitimate azimuthal color gradients in real galaxies . using a synthetic image ,
we have also analyzed the effects of image processing and pixel averaging on the method applied in gg06 and mg09 to detect color gradients and derive pattern speeds .
we have found that pixel averaging ( due to image processing ) systematically decreases the derived @xmath0 , such that it nearly compensates for the systematic effect introduced when neglecting non - circular motions in the analysis .
the net result is that the correct spiral pattern speeds and resonance locations can be obtained .
nevertheless , a residual trend of slightly higher pattern speeds at lower radii can still be discerned ( _ solid _ triangles in fig .
[ myomegas ] and fig .
[ diffom_olr ] ) , so that it is possible to detect the presence of non - circular motions and confirm the link between star formation and disk dynamics .
we have re - examined the results obtained by mg09 . when normalizing the mean radii where gradients were found by the end point radii , in order to treat the whole sample as a single galaxy , we were able to reproduce the trend of @xmath87 with radius expected if non - circular motions are ignored ( as these authors did ) .
the size of the observed @xmath87 in the data only matches the theoretical expectations if spiral patterns end at the olr , and not at the 4:1 resonance ( cf .
[ myomegas ] with both fig .
[ diffom_4_1 ] and fig .
[ diffom_olr ] ) .
future studies of azimuthal color gradients in disk galaxies by orbit calculations should include the effects of disk heating @xcite , and a correct modeling of the imf as a discrete distribution .
these factors may be important , since high mass stars may take different trajectories when compared to low mass stars , due to the effects of dynamical friction @xcite .
we thank the anonymous referee for his / her very constructive suggestions and comments that helped us improve the presentation of this work .
e. martnez - garca acknowledges financial support from cida ( centro de investigaciones de astronoma , venezuela ) .
special thanks to csar briceo , katherine vivas , carlos abad , gustavo bruzual , gladis magris , jess hernndez , and francisco fuenmayor , also at cida .
gcg thanks dgapa ( unam ) grant in-106809 and conacyt grant 50402-f .
ragl acknowledges dgapa ( unam ) grant in111007 and conacyt grant 48589-f .
the differential equations needed to obtain the density and velocity components , product of spiral shocks , are fully explained in @xcite , @xcite , @xcite , and @xcite .
the independent parameters used in this investigation for the solutions are : number of arms , @xmath96 ; spiral arms pitch angle @xmath3 such that @xmath97 ; angular speed of the spiral pattern , @xmath98 km s@xmath22 kpc@xmath22 ; effective speed of sound , @xmath99 km s@xmath22 ; amplitude ratio of the perturbed spiral field to the axisymmetric field , @xmath100 ; rotation velocity of material , .
we also assume that @xmath101constant for the selected range of radii ( 5 - 10 kpc ) . ]
@xmath102 km s@xmath22 . in figure
[ etasp ] , we show the positions of the sonic point and the spiral shock relative to the potential minimum ( @xmath103 ) , with the parameters described above . as already noticed by @xcite and @xcite , the shock front moves away from the potential minimum as the radius increases . in this case , for @xmath104 kpc , the shock front is slightly ahead of the potential minimum .
these relative positions may change for another set of parameters .
the solutions are obtained in curvilinear coordinates @xmath105 ( perpendicular to the spiral equipotential curves ) and @xmath106 ( parallel to the spiral equipotential curves ) .
we adopt the @xmath105 and @xmath106 definitions from @xcite and @xcite .
these definitions differ from those used in @xcite and @xcite by a multiplicative factor @xmath107 , and by an additive constant , chosen such that @xmath103 determines the location of the potential minumum .
the corresponding expressions in terms of the radius , @xmath69 , and the angular coordinate in the rotating frame of the spiral pattern , @xmath108 , are as follows : the @xmath105 coordinate is obtained from the shock solution , together with the velocity components @xmath111 , and @xmath112 .
the subscript 0 labels the unperturbed velocity ( i.e. , in absence of a spiral perturbation ) , and the subscript 1 , the perturbation due to the spiral gravitational field .
the @xmath106 coordinate is obtained by solving the equation : the shock on the subsonic branch occurs at @xmath115 and @xmath116 . @xmath117 and @xmath118 are coordinate values at any point along the streamline ; their maximum possible values are @xmath119 and @xmath120 , i.e. , the point at which the shock occurs on the supersonic branch .. in theory , @xmath121 but , due to a slight nonclosure of the streamlines with radius , the numerical values may differ ( see * ? ? ?
] in order to obtain @xmath116 , we proceed in the way described below .
we define the quantity : @xmath122 where @xmath123 and @xmath124 are the @xmath125 and @xmath70 values obtained when @xmath126 and @xmath127 .
@xmath128 is the average radius of the streamline .
allen , c. , & santillan , a. 1991 , revista mexicana de astronomia y astrofisica , 22 , 255 asiain , r. , figueras , f. , & torra , j. 1999 , , 350 , 434 avedisova , v. s. 1989 , astrophysics , 30 , 83 bash , f. n. , green , e. , & peters , w. l. , iii 1977 , , 217 , 464 bash , f. n. 1979 , , 233 , 524 bash , f. n. 1981 , , 250 , 551 chandrasekhar , s. 1943 , , 97 , 255 charlot , s. , bruzual , g. 2007 , private communication contopoulos , g. , & grosbol , p. 1986
, , 155 , 11 efremov , y. n. 1980 , soviet astronomy letters , 6 , 152 efremov , y. n. 1980 , soviet astronomy letters , 6 , 184 efremov , y. n. 1985 , soviet astron .
lett . , 11 , 69 egusa , f. , kohno , k. , sofue , y. , nakanishi , h. , & komugi , s. 2009 , , 697 , 1870 fernndez , d. , figueras , f. , torra , j. 2008 , , 480 , 735 fujimoto , m. 1968 , iau symposium , 29 , 453 gittins , d. m. , & clarke , c. j. 2004 , , 349 , 909 gmez , g. c. , & martos , m. a. 2009 , chaos in astronomy conference 2007 , astrophysics and space science proceedings , edited by g. contopoulos and p.a
. patsis .
gonzlez , r. a. , & graham , j. r. 1996 , , 460 , 651 ( gg96 ) grosbl , p. , & dottori , h. 2009 , , 499 , l21 heitsch , f. , & hartmann , l. 2008 , , 689 , 290 ishibashi , s. , & yoshii , y. 1984 , , 36 , 41 iye , m. , okamura , s. , hamabe , m. , & watanabe , m. 1982 , , 256 , 103 martnez - garca , e. e. , gonzlez - lpezlira , r. a. , & bruzual - a , g. 2009 , , 694 , 512 martos , m. , hernandez , x. , yez , m. , moreno , e. , & pichardo , b. 2004 , , 350 , l47 navarro , j. f. , frenk , c. s. , & white , s. d. m. 1996 , , 462 , 563 navarro , j. f. , frenk , c. s. , & white , s. d. m. 1997 , , 490 , 493 pichardo , b. , martos , m. , moreno , e. , & espresate , j. 2003 , , 582 , 230 roberts , w. w. 1969 , , 158 , 123 shu , f. h. , milione , v. , gebel , w. , yuan , c. , goldsmith , d. w. , & roberts , w. w. 1972 , , 173 , 557 shu , f. h. , milione , v. , & roberts , w. w. , jr .
1973 , , 183 , 819 saaf , a. f. 1974 , , 189 , 33 sitnik , t. g. 1989 , soviet astron .
lett . , 15 , 388 sitnik , t. g. 1991 , soviet astronomy letters , 17 , 61 stone , j. m. , & norman , m. l. 1992 , , 80 , 753 stone , j. m. , & norman , m. l. 1992 , , 80 , 791 tamburro , d. , rix , h .- w . ,
walter , f. , brinks , e. , de blok , w. j. g. , kennicutt , r. c. , & maclow , m .-
2008 , , 136 , 2872 tosa , m. 1973 , , 25 , 191 vzquez - semadeni , e. , gmez , g. c. , jappsen , a. k. , ballesteros - paredes , j. , gonzlez , r. f. , & klessen , r. s. 2007 , , 657 , 870 wielen , r. 1979 , the large - scale characteristics of the galaxy , 84 , 133 yuan , c. 1969 , , 158 , 889 | assuming that density waves trigger star formation , and that young stars preserve the velocity components of the molecular gas where they are born , we analyze the effects that non - circular gas orbits have on color gradients across spiral arms .
we try two approaches , one involving semi - analytical solutions for spiral shocks , and another with magnetohydrodynamic ( mhd ) numerical simulation data .
we find that , if non - circular motions are ignored , the comparison between observed color gradients and stellar population synthesis models would in principle yield pattern speed values that are systematically too high for regions inside corotation , with the difference between the real and the measured pattern speeds increasing with decreasing radius . on the other hand , image processing and pixel averaging result in systematically lower measured spiral pattern speed values , regardless of the kinematics of stellar orbits .
the net effect is that roughly the correct pattern speeds are recovered , although the trend of higher measured @xmath0 at lower radii ( as expected when non - circular motions exist but are neglected ) should still be observed .
we examine the @xcite photometric data and confirm that this is indeed the case .
the comparison of the size of the systematic pattern speed offset in the data with the predictions of the semi - analytical and mhd models corroborates that spirals are more likely to end at outer lindblad resonance , as these authors had already found . |
figure 1 . comparison of our meson cloud model with data @xcite for @xmath9 .
the solid line is for @xmath54 gev , for which @xmath70 .
the dashed lines are for @xmath56 gev and @xmath57 gev , the range of values constrained by the experimental error of @xmath58 in @xmath59 .
@xmath71 figure 2 .
comparison of our meson cloud model with data @xcite for @xmath68 .
the solid line ( @xmath54 ) is for @xmath72 and @xmath73 gev .
the dashed line shows our result if the @xmath15 cloud contribution is omitted . | we use the meson cloud model of the nucleon to calculate distribution functions for @xmath0 and @xmath1 in the proton . including the effect of the omega meson cloud , with a coupling constant @xmath2 ,
allows a reasonably good description of the data .
# 1@xmath3 # 1 + * omega meson cloud and the proton s light anti - quark distributions * mary alberg@xmath4 , ernest m. henley@xmath5 , and gerald a. miller@xmath6 _ @xmath6 department of physics , university of washington , seattle , wa 98195 , usa _ + _
@xmath7 department of physics , seattle university , seattle , wa 98122 , usa _
+ _ @xmath8 institute for nuclear theory , university of washington , seattle , wa 98195 , usa _
there has been considerable interest in the flavor dependence of the proton s quark distributions .
about ten years ago , the nmc measurement of the integral of the @xmath9 provided convincing evidence for the flavor asymmetry of the sea @xcite . the meson cloud model and the sullivan process were used to explain the momentum fraction distribution of @xmath9 . for reviews ,
see refs . .
interest in the light sea quark distributions was heightened by the recent measurement of the ratio @xmath10 by means of the drell - yan process @xcite .
similar results for @xmath9 were obtained by hermes @xcite . in the context of previous work on the light antiquark asymmetry in the nucleon , a good description of the data was obtained for the @xmath11-dependence ( momentum fraction ) of @xmath9 , but not for the ratio of @xmath10 @xcite .
the meson cloud models provide ratios that either increase monotonically with @xmath11 or turn back towards unity too slowly @xcite .
various explanations have been advanced for the discrepancy , e.g. , effects of the @xmath12 @xcite-@xcite , and the influence of the pauli exclusion principle @xcite , adjustment of parameters @xcite , but none of these changes provides a satisfactory description of the @xmath10 distribution function .
the use of a soft @xmath13n@xmath12 form factor enables a reasonable description of @xmath14 @xcite , but then @xmath10 is too large .
the importance of understanding the ratio @xmath10 was emphasized recently @xcite . in this letter
we show that the inclusion of the @xmath15 meson along with pion cloud effects , provides a good description of the present data for both @xmath9 and @xmath10 .
the importance of the @xmath15 in providing the repulsive short - range part of the nucleon - nucleon and consequently attractive nucleon - antinucleon force has been known for a long time .
the compton wavelength of this meson is small , but the large ( almost puzzlingly large @xmath16 ) coupling constants universally used to describe the nucleon - nucleon scattering data cause the potential to be non - vanishing over a fairly large range .
the @xmath15 meson also may be very important in deep inelastic scattering from nuclei @xcite , so it is natural to consider its effects for a proton target .
we provide the usual formulae for the effects of the meson cloud ( generalized sullivan process ) in order to facilitate the reader s understanding of what follows .
the wave function of the proton is written in terms of fock states with and without mesons @xmath17 here z is a wavefunction renormalization constant , @xmath18 is the probability amplitude for finding a physical nucleon in a state consisting of a baryon , @xmath19 with longitudinal momentum fraction y and meson @xmath20 of momentum fraction ( 1-y ) and squared transverse relative momentum @xmath21 .
the quark distribution function @xmath22 of a proton is given by @xmath23 with @xmath24 @xmath25 and @xmath26 the cut - offs required in the model are taken from ref .
@xmath27 where @xmath28 is a cut - off parameter for each meson and @xmath29 and @xmath30 are the usual kinematical variables , expressed in terms of @xmath31 and @xmath32 . such a form is required to respect the identity ( [ sym ] ) .
the expressions for the splitting functions @xmath33 are those given by as derived in ref .
@xcite .
we include specifically @xmath13 , @xmath15 , and @xmath34 mesons , but the latter is negligible . in the present paper , we omit the effects of the @xmath35 meson as well as those of the intermediate @xmath12 . the former increases @xmath9 , whereas the latter decreases it , so these effects tend to cancel .
these effects have been included by previous authors , and do not provide a satisfactory description of the ratio @xmath36 .
the number of each type of meson , @xmath37 , is obtained by integrating the square of @xmath33 over @xmath32 .
then for us @xmath38 .
we need to discuss the functions @xmath39 and @xmath40 of our calculation .
those for the nucleon and pion are measured , but the quark distribution functions of the @xmath15 meson are unknown .
the bag model suggests that the structure functions of the @xmath15 , @xmath41 and @xmath13 mesons are the same .
the near equality is more believable for the @xmath15 and @xmath35 , than for that between the vector and pseudoscalar mesons .
however , it has been traditional @xcite to assume that the structure function of the @xmath35 and @xmath13 are the same .
thus we use @xcite @xmath42 for the valence and sea quark distributions of both the pion and omega mesons .
the bare nucleon sea is parametrized @xcite as @xmath43 the value of @xmath44 gev is chosen to reproduce the range of allowed values of the integral @xcite @xmath45 using the sum rule of henley & miller @xcite : @xmath46 .
the parameter @xmath47 is expected to be larger than @xmath48 @xcite , and we used the range : 1.3 gev @xmath49 1.8 gev .
the results of our calculations for @xmath50 are shown in fig . 1 , and those for the ratio @xmath51 are shown in fig . 2 .
the @xmath13-nucleon coupling constant is taken as @xmath52 , the @xmath15-proton coupling constant is taken to be @xmath53 = 8.1 . in fig . 1
the solid line is for @xmath54 gev , for which @xmath55 .
the dashed lines are for @xmath56 gev and @xmath57 gev , the range of values constrained by the experimental error of @xmath58 in @xmath59 .
the @xmath60 meson , like any other isoscalar meson , has no effect here , so that this curve is that from the pion alone . in fig .
2 the solid curve shown is for @xmath61 gev . the dashed line shows the effect of leaving out the @xmath15 cloud contribution .
we have examined the effect of varying both @xmath47 and @xmath62 in the range @xmath63 gev @xmath64 gev and @xmath65 .
the larger values of @xmath62 are favored in fits to nucleon - nucleon scattering data using one - boson exchange potentials @xcite and the smaller ones from dispersion relation descriptions of forward nucleon - nucleon scattering .
the larger values of @xmath62 and @xmath66 tend to give too small values of the ratio @xmath67 , while decreasing @xmath62 or @xmath66 causes the maximum value of @xmath10 to be too large and to appear at too high a value of @xmath11 .
the effects of @xmath34 mesons are not included in the curves shown in the figures .
these provide only a 1 to 2% change in @xmath68 for phenomenologically viable coupling constants .
although we also examined the change of distribution function for the bare quarks suggested by @xcite , the effect is sufficiently small when added to that of the @xmath15 meson that we do not show it .
it is clear from the figures that a good description of the present data is provided by the inclusion of the @xmath15 meson with a reasonable coupling constant , and that therefore the meson cloud picture can be successful . as mentioned above
there are a number of effects discussed in the literature , not included here , which contribute to the proton sea .
in addition , one could include the effects of the @xmath69 meson , which would also tend to suppress @xmath68 . including these effects is likely to improve the description of the data or modify the parameters describing the @xmath15-nucleon interaction .
the existence of better data would provide a severe test of the present model , and the prospects of such seem imminent @xcite . but it is fair to conclude that the use of the @xmath15 along with the previously suggested meson cloud effects does allow for a good description of the present data .
we thank the usdoe and the usnsf for partial support of this work .
we thank r.j .
holt for a useful discussion .
one of us ( emh ) thanks j. speth for a useful discussion and for his hospitality at the kfz in juelich .
99 p. amaudruz et al .
, nmc collaboration , phys .
rev .
lett .
* 66 * ( 1991 ) 2712 .
j. speth and a.w .
thomas , _ advances in nuclear physics _ , vol .
24 , ed .
j.w .
negele and e.w .
vogt ( plenum press , ny ) 1998 .
s. kumano , phys . rep .
* 303 * ( 1998 ) 183 .
a. baldit et al . , phys . lett . * 332 * ( 1994 ) 244 .
e866 collaboration , e.a .
hawker et al .
, phys .
rev .
lett .
* 80 * ( 1998 ) 3715 ; j.c .
peng et al . , phys .
rev * d58 * ( 1999 ) 092004 .
k. ackerstaff , et al .
, phys .
rev .
lett .
* 81 * ( 1998 ) 5519 .
m. alberg , t. falter and e. m. henley , nucl . phys .
* a644 * ( 1998 ) 93 .
s. kumano , phys . rev .
* d43 * ( 1991 ) 3067 .
a. signal , a.w .
schreiber and a.w .
thomas , mod .
. lett .
* a6 * ( 1991)271 .
w. koepf , l.l .
frankfurt and m. strikman , phys . rev .
* d53 * ( 1996 ) 2586 .
w. melnitchouk , j. speth , and a.w .
thomas , phys . rev .
* d59 * ( 1998 ) 014033 .
n.n .
nikolaev , w. schfer , a. szczurek and j. speth , phys . rev .
* d 60 * ( 1999 ) 014004 . a.w .
thomas , summary talk at the 1999 panic conference ( to be published , nucl .
phys .
a ) .
g.a .
miller , phys . rev .
* c56 * ( 1997 ) r8 ; phys . rev .
* c56 * ( 1997 ) 2789 .
a. szczurek , m. ericson , h. holtmann and j. speth , nucl . phys .
* a596 * ( 1996 ) 397 .
h. holtmann , ( mesonen i m nukleon und ihre auswirkungen in elastischer und tiefinelastischer streung ) , forschungszentrum juelich , dissertation , university of bonn ( 1995 ) ch .
elster , k. holinde , and r machleidt phys . rep . * 149 * ( 1987 ) 1 .
e.m .
henley and g.a .
miller , phys .
lett . *
b251 * ( 1990 ) 453 . w.y .-
p .
hwang , j. speth and g.e .
brown , z. phys . * a339 * ( 1991 ) 383 ; a. szczurek and j. speth , nucl . phys .
* a555 * ( 1993 ) 249 .
p.j .
sutton , a.d .
martin , r.g .
roberts and w.j .
stirling , phys .
rev .
* d45 * ( 1992 ) 2349 . c. gagliardi ( private communication and to be published ) |
recent observations indicate that the universe is accelerating , and it is spatially flat ( @xmath1 ) .
approximately @xmath2 of the total energy density of the universe @xmath3 g/@xmath4 consists of ordinary matter ( @xmath5 ) , and @xmath6 of the energy density corresponds to dark energy ( @xmath7 ) @xcite .
there are three basic scenarios describing the evolution of the universe filled by dark energy .
\i ) from a purely phenomenological point of view , the simplest possibility is that the dark energy @xmath8 is represented by a positive vacuum energy ( cosmological constant ) @xmath9 .
if this is the case , the universe will reach de sitter ( ds ) regime and expand exponentially for an indefinitely long time , @xmath10 .
this possibility seems much more natural than the other two possibility to be discussed below .
\ii ) it may also happen that the dark energy is the energy of a slowly changing scalar field @xmath11 with equation of state @xmath12 , @xmath13 .
in most of the models of dark energy it is assumed that the cosmological constant is equal to zero , and the potential energy @xmath14 of the scalar field driving the present stage of acceleration , slowly decreases and eventually vanishes as the field rolls to @xmath15 , see e.g. @xcite . in this case , after a transient ds - like stage , the speed of expansion of the universe decreases , and the universe reaches minkowski regime .
\iii ) it is also possible that @xmath14 has a minimum at @xmath16 , or that it does not have any minimum at all and the field @xmath11 is free to fall to @xmath17 . in this case _ the universe eventually collapses , even if it is flat _
the simplest way to understand this unusual effect is to analyse the friedmann equation @xmath18 ( in units @xmath19 ) .
the positive energy density of a normal matter , as well as the positive kinetic energy density of the scalar field , tend to decrease in an expanding universe . at some moment
, the total energy density @xmath20 , including the negative contribution @xmath16 , vanishes .
once it happens , the universe , in accordance with the equation @xmath18 , stops expanding and enters the stage of irreversible collapse . the last possibility for
a while did not attract much attention .
there was no specific reason to expect that the present regime of acceleration is going to end , and there was even less reason to believe that the universe is going to collapse any time soon .
unfortunately , despite many attempts , we were unable to obtain a good theoretical description of the models i ) and ii ) in the context of m - theory .
we will discuss this issue in sect .
2 . meanwhile , in @xcite it was found that one can describe the present state of acceleration of the universe in a broad class of models based on n=8 extended supergravity ( the theory closely related to m - theory ) .
however , the universe described by these models typically collapses within the time @xmath21 comparable to its present age @xmath22 billion years . in the beginning , this seemed to be a model - specific result that should be taken seriously only if one can construct fully realistic models of elementary particles based on extended supergravity .
however , this result is valid not only in for the models based on extended supergravity but for many phenomenological models of dark energy based on the minimal n=1 supergravity @xcite . in this paper
we will argue that this result is even more general .
it can be obtained in almost every model of dark energy , either based on supergravity or not , if one takes into account the possibility that the effective potential of the field @xmath11 may have a minimum at @xmath16 or may be unbounded from below .
we will first review the possibilities to describe the accelerating universe starting with m / string theory and supergravity .
then we will describe our general argument .
the standard approach to the description of our world in m / string - theory is based on the assumption that our space - time is 11 or 10 dimensional , but 7 or 6 of these dimensions are compactified .
the finitness of the volume of the compactified space is required so that the original d=11 or d=10 theory can be related to the resulting d=4 theory .
one can describe the scale of compactification for example , in string theory by introducing scalar moduli field @xmath11 , which appears as a coefficient @xmath23 in front of the potential energy @xmath24 in 4d .
( we use the units @xmath25 , where @xmath26 . ) here @xmath24 is the value of the potential at which other scalar fields of the higher - dimensional theory are stabilized .
however , the mechanism of stabilization of the compactified space is still unknown . in the absence of such mechanism
, the term @xmath27 leads to the runaway behaviour @xmath28 , which implies _
decompactification_. during this process , the energy density @xmath27 falls down very quickly .
a similar result is valid if we consider d=10 string theory as a result of compactification of d=11 supergravity .
the bottom line is : _ during the cosmological evolution in d=4 the runaway moduli represent decompactification of all internal dimensions of m - theory_. the scale factor of the universe @xmath29 in the theories with exponential potentials of the type of @xmath30 grows as @xmath31 . in
order to describe acceleration of the universe in such theories one would need to have @xmath32 .
in the theory with the potential @xmath33 the universe can only decelerate . in application to string cosmology these observations
imply that until we learn how to stabilize the compactified space , we can not describe the accelerating universe approaching ds regime , as well as the accelerating universe with the energy density slowly approaching zero .
one way to avoid the problems discussed above is to consider the models with a non - compact internal space @xcite .
one may start with d=11 or d=10 supergravity with internal space with an infinite volume and relate it to n=8 supergravity in d=4 @xcite .
this is called ` non - compactification . ' in this approach the connection between the original d=10 or d=11 theory and our d=4 world is more complicated than in the usual case of dimensional reduction . however , one can study these theories directly in d=4 .
these theories are interesting because they have maximal amount of supersymmetry , related to d=11 and/or d=10 supergravities with non - compact internal spaces .
the number of such models successfully describing an accelerated universe is very limited , due to the maximal amount of supersymmetries .
some of these theories have ds solutions and can describe dark energy @xcite .
these ds solutions correspond to the extrema of the effective potentials @xmath14 for some scalar fields @xmath11 .
an interesting and very unusual feature of these scalars in all known theories with @xmath34 is that their mass squared is quantized in units of the hubble constant @xmath35 corresponding to ds solutions : @xmath36 , where @xmath37 are some integers of the order 1 .
this property was first observed in @xcite for a large class of extended supergravities with unstable ds vacua , and confirmed and discussed in detail more recently in @xcite with respect to a new class of @xmath38 gauged supergravities with stable ds vacua @xcite .
the meaning of this result can be explained in the following way .
the simplest potential for a scalar field @xmath11 in this theory has the form [ simplepot0 ] v ( ) = ( 2 - 2 ) . usually the potential near its extremum can be represented as @xmath39 , where @xmath40 and @xmath41 are two free independent parameters . however , in extended supergravities with @xmath42 one always has @xmath43 , where @xmath37 are integers ( we are using units @xmath25 ) @xcite . taking into account that in ds space the hubble constant is given by @xmath44 ,
one has , for @xmath45 , @xmath46 .
in particular , in all known versions of @xmath47 supergravity ds vacuum corresponds to an unstable maximum , @xmath48 @xcite , _ i.e. _ at @xmath45 one has [ simplepot ] v ( ) = ( 1 -^2 ) = 3h^2(1 -^2 ) .
one can easily verify that the simplest potential ( [ simplepot0 ] ) satisfies this rule .
the main property of this potential is that @xmath49 .
one can show that a homogeneous field @xmath50 with @xmath51 in the universe with the hubble constant @xmath35 grows as follows : @xmath52 , where @xmath53 . consequently , in the universe with the energy density dominated by @xmath14 it takes time @xmath54 until the scalar field rolls down from @xmath55 to the region @xmath56 , where @xmath14 becomes negative .
once it happens , the universe rapidly collapses .
= 5.3 cm note , that the present age of the universe @xmath57 is approximately equal to @xmath58 , and the total time of the development of the instability leading to the global collapse of the universe is given by @xmath54 , which is also of the same order as @xmath35 , unless @xmath55 is exponentially small .
this explains the main result of ref .
@xcite : the universe described by this class of theories is going to collapse within the time @xmath21 comparable to its present age @xmath22 billion years , see fig .
[ scalefactorcoll ] .
there are also 3 different cosmological models of dark energy based on n=8 supergravity , with the following potentials of two fields , @xmath11 and @xmath59 .
we will list them here , leaving a detailed description of their cosmological implications for the future publication @xcite [ 332 ] v ( , ) = e^- ( 3- ( 2 ) ) .
this is an improved version of townsend s model of m - theory quintessence @xcite .
townsend used this potential at @xmath60 and found the exponential potential @xmath61 which may describe the current acceleration with @xmath62 .
the complete form of the potential shows that the full potential of this theory is unbounded from below and unstable with respect to the generation of the field @xmath63 .
the speed of the development of the instability is determined by the curvature of the potential in the @xmath11-direction , @xmath64 , i.e. @xmath65 .
this allows for the existence of a stage of accelerated expansion , but eventually the universe collapses , just like in the theory ( [ simplepot0 ] ) discussed above .
the second model is [ 332a ] v ( ) = e^- ( 24 e^- -8 e^2- 3 e^-8 ) .
the @xmath11 part of the potential has a ds maximum at @xmath66 .
it may describe the current acceleration during the slow growth of the field @xmath67 : @xmath68 . the instability with respect to the field @xmath11 eventually develops and the universe collapses . the last model of this type has a saddle point ds solution at @xmath69 [ 332b ] v ( ) = ( 2- 2 ) . at the point @xmath70 this model reduces to the model ( [ simplepot0 ] ) which describes dark energy and eventual collapse of the universe .
a lot of work should be done to incorporate usual matter fields and construct realistic cosmology in n=8 supergravity .
however , it is quite encouraging that there is a class of models based on n=8 supergravity which can describe the present stage of acceleration of the universe .
all of these models share the same property : at some moment expansion of the universe stops and the universe collapses within time @xmath21 comparable to its present age @xmath22 billion years .
the main reason for the coincidence of the two different time scales , the present age of the universe and the time until the big crunch , is the relation @xmath71 . but this relation appears not only in the extended supergravity .
it is often valid for the moduli fields in @xmath72 supergravity @xcite . in particular , it is valid in the simplest polnyi - type toy model for dark energy in @xmath72 supergravity @xcite .
if one does not fine - tune the value of the cosmological constant in this model to be equal to zero , one has two equally compelling options : @xmath73 and @xmath16 in the minimum of the effective potential . in the first case
, the universe enters ds regime of eternal expansion . in the second case
, the universe collapses .
the time until the global collapse depends on the parameters of the model and the initial conditions , but typically it is of the same order as @xmath22 billion years , just as in the extended supergravity @xcite .
another interesting model is the axion quintessence @xcite . in the m - theory motivated
version of this model proposed in @xcite one has @xmath74 , where the value of constant @xmath75 depends on the details of the model . for @xmath76 , @xmath77 one finds @xmath78 . according to @xcite ,
this version of the axion quintessence model can successfully describe the present stage of acceleration of the universe , but , just like the @xmath47 models , it leads to a global collapse of the universe in the future within the typical time @xmath79 billion years .
in fact , the crucial relation @xmath80 is valid in most of the dark energy models .
it is the standard inflationary slow - roll condition , which should be valid at the present stage of the late - time inflation / acceleration of the universe .
sometimes this slow - roll condition can be violated @xcite , but typically one can not have a prolonged stage of acceleration of the universe for @xmath81 . as a result , one can give a simple argument suggesting that the coincidence of the two different time scales , the present age of the universe and the time remaining until the big crunch , is a generic property of many models of dark energy .
let us make the simplest assumption that the expansion of the universe at @xmath82 can be approximately described by the simple power - law equation @xmath83 where @xmath84 is some constant .
the hubble constant in the universe with @xmath85 is given by @xmath86 , which means that the total energy density is equal to @xmath87 .
the parameter @xmath84 is determined by the continuity condition for @xmath88 : @xmath89 the last part of this equation is the observed relation between the present value of @xmath88 and the age of the universe in the simplest @xmath40cdm model .
this gives @xmath90 .
the energy density will become @xmath91 times smaller than now at the time @xmath92 .
in particular , the energy density will become 2 times smaller at the time @xmath93 from now , and it will become 9 times smaller at @xmath94 .
how large is this time interval ?
acceleration of the universe implies that @xmath95 .
taking for definiteness @xmath96 , one finds that the density of the universe will decrease 2 times at the moment @xmath97 billion years from now , and it will drop 9 times @xmath98 billion years from now .
now let us take any model of dark energy with @xmath99 and add to the scalar potential @xmath14 a tiny negative cosmological constant @xmath100 with @xmath101 .
the evolution of the universe up to the present moment will not change significantly .
one can make this model even better by multiplying @xmath14 by some factor @xmath102 to ensure that the present value of @xmath103 remains equal @xmath6 even after we add a negative constant @xmath100 to @xmath14 .
this model will remain a viable model of dark energy .
however , after some time @xmath104 the value of @xmath14 will drop down more than n times , the total energy density of the universe will vanish , the universe will stop expanding , and soon after that it will collapse . if @xmath105 is not too large , the universe collapses at the time @xmath106 .
one could think that the argument given above works only for the marginal situation , when @xmath107 with @xmath108 .
however , the final result is very general .
our procedure of modification of the potential ( subtraction of a constant and a subsequent compensation of the decrease of @xmath103 via the multiplication of @xmath14 by a constant @xmath102 ) works until the moment when the theory no longer represents dark energy because the potential becomes too steep .
this happens for any theory of dark energy , even if the original potential was extremely flat ( which corresponds to @xmath109 ) .
once it happens , the field rolls down within the time @xmath110 .
the rolling field rapidly approaches the region of negative @xmath14 , which typically leads to the collapse of the universe within the time @xmath110 .
this suggests that many models of dark energy considered in the literature have viable counterparts that can be obtained from the original models by adding a negative cosmological constant @xmath111 .
unless the absolute value of this extra term is many orders of magnitude smaller than @xmath112 , these models will describe the universe collapsing within the time comparable to the present age of the universe .
as an example illustrating the general argument given in the previous section , let us consider dark energy described by the scalar field @xmath11 with an exponential potential @xmath113 we already mentioned that this theory describes an accelerating universe for @xmath32 .
however , in the realistic situation this condition can be slightly relaxed .
one should take into account that the post - inflationary universe rapidly expanded , being dominated by hot matter and then by cold dark matter .
this expansion does not allow the field @xmath11 to move until the energy density of matter becomes sufficiently small .
therefore in the beginning the kinetic energy of the scalar field is very small , so it has the same equation of state as the cosmological constant .
then the field starts moving slowly , loosing its energy at a much slower paste than cdm .
eventually , the universe enters the stage with @xmath114 ( the present time ) , and then @xmath103 continues to grow . as a result , one can have @xmath115 and @xmath116 at present for @xmath117 @xcite . in our investigation we will assume , without loosing the generality , that the initial value of the field was @xmath118 ( one can always rescale the field and the potential )
. then one should find such value of the parameter @xmath40 that the universe enters the stage with @xmath114 at the same time when its hubble constant acquires its present value @xmath119 .
this requires fine - tuning , but this is the same fine - tuning that hampers all models of dark energy @xcite .
now we will consider a class of potentials of a more general type , containing a constant negative contribution , as suggested in the previous section : @xmath120 here @xmath75 is some positive constant .
the constant @xmath121 should be found anew for each new value of @xmath75 , just as we did for the case @xmath76 described above .
this can be done , and the cosmological evolution of this model can be easily studied using the methods described in @xcite . here
we will only present the results of our investigation of a model with @xmath122 .
= 5.3 cm figure [ scalefactorcoll2 ] shows the evolution of the scale factor of the universe for 4 different values of parameter @xmath75 : @xmath123 and @xmath124 . for all of these cases one can find such parameters @xmath125 that the value of the hubble constant at @xmath126 coincides with its present value ( the curves @xmath29 have the same derivative at @xmath127 in figure [ scalefactorcoll2 ] ) .
= 5.3 cm all of the models with @xmath123 with the potentials shown in figure [ pot ] can represent dark energy with @xmath115 at the present moment ( the point @xmath128 in figure [ omegacoll ] ) .
however , the largest value of @xmath103 in the case @xmath129 is @xmath130 , and the equation of state @xmath131 in this case blows up near @xmath128 , see figure [ wcoll ] .
= 5.7 cm as we see , the universe with @xmath76 ( thick red line in figures [ pot]-[acc ] ) always continues its accelerated expansion .
however , in all other cases the universe collapses within the time ranging from @xmath132 billion years ( for @xmath129 ) to @xmath133 billion years ( for @xmath134 ) , in agreement with the argument given in the previous section .
the universe with @xmath129 never accelerates , so this model is ruled out by the existing observations .
thus , for every model @xmath135 successfully describing dark energy there exist many other models @xmath136 with @xmath137 which provide an equally good description of the present stage of acceleration , but lead to the global collapse of the universe within the next @xmath138 years .
similar results can be obtained for many other models of dark energy , e.g. for the theories @xmath139 with all possible values of @xmath140 , or for the models with the inverse power law potentials .
recent discovery of acceleration of the universe is one of the major challenges for the modern theory of fundamental interactions .
after many attempts to explain why the cosmological constant must be zero , the theorists switched to the new paradigm and started trying to explain why it should be positive and why , consequently , the universe should expand forever .
the first attempts to do so in the context of m / string theory and extended supergravity revealed many problems described , e.g. , in @xcite
. then we learned that one can describe acceleration of the universe in extended supergravity , but in all models based on extended supergravity , with the exception of the n=2 model of @xcite , the regime of acceleration is unstable @xcite .
typically it ends by a global collapse of the universe within the time comparable with the present age of the universe , @xmath141 billion years @xcite . in this paper
we have shown that the possibility of a global collapse is not specific to supergravity but is , in fact , quite generic .
for every model of dark energy describing eternally expanding universe one can construct many closely related models which describe the present stage of acceleration of the universe followed by its global collapse .
this does not mean that we are making a doomsday prediction .
none of the existing theoretical models of dark energy look particularly natural and attractive .
it may happen that eventually we will find good theoretical models describing an eternally accelerating universe , or conclude , on the basis of anthropic considerations , that dark energy should change in time extremely slowly , so that the collapse will occur exponentially far away in the future @xcite .
we hope to return to this question in the future publications .
however , in the absence of a compelling theory of dark energy one may also consider a more humble approach and try to compare predictions of various models of dark energy with observations , see e.g. @xcite .
if the universe is going to collapse , then in the beginning of this process the speed of expansion of the universe should gradually slow down , see fig .
this is accompanied by the rapid growth of the parameter @xmath131 at small @xmath142 , as shown in fig .
[ wcoll ] .
we find it quite significant that some of the models predicting global collapse can be already ruled out by the existing observational data .
for example , all models @xmath143 predicting the global collapse within the next 18 billion years , do not describe the present stage of acceleration , and therefore contradict the recent cosmological observations .
thus , even though the observational data can not rule out the general possibility of the global collapse in the distant future , they can help us to put strong constraints on the time of the possible big crunch .
it is a pleasure to thank r. bond , l. kofman , j. kratochvil , e. linder , s. prokushkin and m. shmakova , for useful discussions .
this work was supported by nsf grant phy-9870115 .
the work by a.l . was also supported by the templeton foundation grant no . 938-cos273 .
s. perlmutter _ et al .
_ , `` measurements of omega and lambda from 42 high - redshift supernovae , '' astrophys.j . * 517 * , 565 ( 1999 ) [ astro - ph/9812133 ] , see also http://snap.lbl.gov ; a. g. riess _ et al .
_ , `` observational evidence from supernovae for an accelerating universe and a cosmological constant , '' astron .
j. * 116 * , 1009 ( 1998 ) [ astro - ph/9805201 ] .
j. l. sievers _ et al . _ , `` cosmological parameters from cosmic background imager observations and comparisons with boomerang , dasi , and maxima , '' astro - ph/0205387 ; j. r. bond _ et al . _ , `` the cosmic microwave background and inflation , then and now , '' arxiv : astro - ph/0210007 . a. d. dolgov , `` an attempt to get rid of the cosmological constant , '' in : _ the very early universe _ ,
gibbons , s.w . hawking and s.siklos ( cambridge university press 1983 ) , pp . 449 - 458 ; c. wetterich , `` cosmology and the fate of dilatation symmetry , '' nucl . phys .
b * 302 * , 668 ( 1988 ) ; p. g. ferreira and m. joyce , `` cosmology with a primordial scaling field , '' phys .
d * 58 * , 023503 ( 1998 ) [ arxiv : astro - ph/9711102 ] ; b. ratra and p. j. peebles , `` cosmological consequences of a rolling homogeneous scalar field , '' phys .
d * 37 * , 3406 ( 1988 ) ; i. zlatev , l. m. wang and p. j. steinhardt , `` quintessence , cosmic coincidence , and the cosmological constant , '' phys .
lett . * 82 * , 896 ( 1999 ) [ arxiv : astro - ph/9807002 ] .
linde , `` inflation and quantum cosmology , '' print-86 - 0888 ( june 1986 ) , in : _ three hundred years of gravitation _ , ( eds . : hawking , s.w . and
israel , w. , cambridge univ . press , 1987 ) , 604 - 630 ; l. m. krauss and m. s. turner , `` geometry and destiny , '' gen.rel . grav .
* 31 * , 1453 ( 1999 ) [ arxiv : astro - ph/9904020 ] ; a. a. starobinsky , `` future and origin of our universe : modern view , '' grav .
cosmol .
* 6 * , 157 ( 2000 ) [ arxiv : astro - ph/9912054 ] ; g. n. felder , a. frolov , l. kofman and a. linde , `` cosmology with negative potentials , '' phys .
d * 66 * , 023507 ( 2002 ) [ arxiv : hep - th/0202017 ] .
r. kallosh , a. linde , s. prokushkin and m. shmakova , `` supergravity , dark energy and the fate of the universe , '' phys .
d * 66 * , 123503 ( 2002 ) [ arxiv : hep - th/0208156 ] .
r. kallosh and a. linde , `` m - theory , cosmological constant and anthropic principle , '' arxiv : hep - th/0208157 . c. m. hull and n. p. warner ,
`` noncompact gaugings from higher dimensions , '' class . quant .
grav . * 5 * , 1517 ( 1988 ) .
p. fre , m. trigiante and a. van proeyen , `` stable de sitter vacua from n = 2 supergravity , '' class .
* 19 * , 4167 ( 2002 ) [ arxiv : hep - th/0205119 ] .
m. dine , w. fischler and d. nemeschansky , `` solution of the entropy crisis of supersymmetric theories , '' phys .
b * 136 * , 169 ( 1984 ) ; g. d. coughlan , r. holman , p. ramond and g. g. ross , `` supersymmetry and the entropy crisis , '' phys.lett .
b * 140 * , 44 ( 1984 ) ; m. dine , l. randall and s. thomas , `` baryogenesis from flat directions of the supersymmetric standard model , '' nucl .
b * 458 * , 291 ( 1996 ) [ arxiv : hep - ph/9507453 ] ; a. d. linde , `` relaxing the cosmological moduli problem , '' phys .
d * 53 * , 4129 ( 1996 ) [ arxiv : hep - th/9601083 ] .
j. a. frieman , c. t. hill , a. stebbins and i. waga , `` cosmology with ultralight pseudo nambu - goldstone bosons , '' phys .
lett . * 75 * , 2077 ( 1995 ) [ arxiv : astro - ph/9505060 ] ; i. waga and j. a. frieman , `` new constraints from high redshift supernovae and lensing statistics upon scalar field cosmologies , '' phys .
d * 62 * , 043521 ( 2000 ) [ arxiv : astro - ph/0001354 ]
. k. choi , `` string or m theory axion as a quintessence , '' phys.rev .
d * 62 * , 043509 ( 2000 ) [ arxiv : hep - ph/9902292 ] . g. w. gibbons , `` apects of supergravity theories , '' in _ supersymmetry , supergravity and related topics _ , eds
. f. del aguila , j.a .
de azcrraga and l.e .
ibaez ( world scientific 1985 ) pp . 346 - 351 ; j. maldacena and c. nuez , `` supergravity description of field theories on curved manifolds and a no - go theorem , '' int .
a16 * ( 2001 ) 822 .
s. hellerman , n. kaloper and l. susskind , `` string theory and quintessence , '' jhep * 0106 * , 003 ( 2001 ) [ arxiv : hep - th/0104180 ] ; w. fischler , a. kashani - poor , r. mcnees and s. paban , `` the acceleration of the universe , a challenge for string theory , '' jhep * 0107 * , 003 ( 2001 ) [ arxiv : hep - th/0104181 ] .
j. garriga and a. vilenkin , `` testable anthropic predictions for dark energy , '' arxiv : astro - ph/0210358 .
p. s. corasaniti and e. j. copeland , `` constraining the quintessence equation of state with snia data and cmb peaks , '' phys .
d * 65 * , 043004 ( 2002 ) [ arxiv : astro - ph/0107378 ] ; j. weller and a. albrecht , `` future supernovae observations as a probe of dark energy , '' phys .
d * 65 * , 103512 ( 2002 ) [ arxiv : astro - ph/0106079 ] ; i. maor , r. brustein , j. mcmahon and p. j. steinhardt , `` measuring the equation - of - state of the universe : pitfalls and prospects , '' phys .
d * 65 * , 123003 ( 2002 ) [ arxiv : astro - ph/0112526 ] ; w. hu , `` dark energy and matter evolution from lensing tomography , '' arxiv : astro - ph/0208093 ; j. a. frieman , d. huterer , e. v. linder and m. s. turner , `` probing dark energy with supernovae : exploiting complementarity with the cosmic microwave background , '' arxiv : astro - ph/0208100 ; j. j. mohr , b. oshea , a. e. evrard , j. bialek and z. haiman , `` studying dark energy with galaxy cluster surveys , '' arxiv : astro - ph/0208102 . | it is often assumed that in the course of the evolution of the universe , the dark energy either vanishes or becomes a positive constant .
however , recently it was shown that in many models based on supergravity , the dark energy eventually becomes negative and the universe collapses within the time comparable to the present age of the universe .
we will show that this conclusion is not limited to the models based on supergravity : in many models describing the present stage of acceleration of the universe , the dark energy eventually becomes negative , which triggers the collapse of the universe within the time @xmath0 years .
the theories of this type have certain distinguishing features that can be tested by cosmological observations . |
the broadband spectral energy distribution ( sed ) of active galactic nuclei ( agn ) has been studied for many years . the unobservable gap between the extreme uv and soft x - rays , imposed by galactic photoelectric absorption sets a major barrier to reconstructing the overall sed .
this is particularly problematic because it is in this energy range that the peak of the sed often occurs . due to this difficulty , most previous work focused on spectral properties of either the optical / uv side ( e.g. sanders et al .
1989 ; francis et al . 1991 ; zheng et al . 1997 ; vanden berk et al .
2001 ) , or the x - ray side ( e.g. wilkes & elvis 1987 ; green et al . 1995 ; george et al .
however , the multi - wavelength combination is required to see the overall sed behaviour ( e.g. elvis et al .
1994 ; laor et al .
1997 ; richards et al . 2006 ;
shang et al .
2005 ; lusso et al .
2010 , hereafter : lusso10 ) . in order to recover the spectral shape in the uv / soft x - ray gap ,
some studies simply connected the optical / uv and x - ray spectra to estimate the sed ( e.g. elvis et al .
1994 ; richards et al . 2006 ; shang et al .
2005 ; lusso10 ) .
as the origin of the optical and hard x - ray emission became clearer , more broadband sed models with multiple physical components were used to fit the multi - waveband spectra .
these were used to extrapolate across the uv gap based on the model assumptions , but with better justification than before ( e.g. middleton , done & gierliski 2007 ; vasudevan & fabian 2007 , hereafter : vf07 ; vasudevan & fabian 2009 , hereafter : vf09 ; jin et al .
2009 ; grupe et al .
2010 , hereafter : grupe10 ) .
for example , vf07 used a power law plus accretion disc model to construct the broadband sed for their sample .
this technique was repeated in vf09 based on 29 agns from the reverberation mapping sample ( peterson et al .
2004 ) , and using simultaneous optical / uv and x - ray data from xmm - newton .
grupe10 used an absorbed power law for the x - ray spectrum , and an exponentially cut - off power law ( or a broken power law ) for the optical / uv spectrum , to construct the broadband sed for their sample , for which simultaneous optical , uv , x - ray data are available from swift .
further discussion of some previous sed studies is given in jin et al .
( 2011 ) ( hereafter : paper - i ) . within these works ,
some specific agn parameters were studied , including black hole mass ( m@xmath6 ) , eddington ratio ( @xmath16 ) , bolometric luminosity ( l@xmath15 ) , 2 - 10 kev luminosity ( l@xmath17 ) , 2 - 10 kev photon index ( @xmath0 ) , h@xmath18 fwhm . also , some parameters directly related to the broadband sed shape were proposed , such as the optical to x - ray spectral index ( @xmath7 , assuming @xmath19 to ensure a non - negative @xmath7 . ] , tananbaum et al .
1979 ) , 2 - 10 kev bolometric correction ( @xmath14 , defined as l@xmath15/l@xmath17 , e.g. vf07 ) , 5100 luminosity scaling factor ( @xmath12 , defined as l@xmath15/l@xmath20 , where l@xmath20 is the monochromatic luminosity at 5100 , kaspi et al . 2000 ) .
since the number of agn with both high quality optical / uv and x - ray spectra is relatively small , much effort is devoted to search for correlations among the key sed parameters , especially those parameters capable of representing the broadband sed shape .
then , for those many agns lacking sufficient spectral information , these correlations can be used to predict the seds that can not be defined from direct observation .
indeed , many such parameter correlations have been proposed .
for example , vf07 reported a strong correlation between @xmath14 and @xmath13 ( vf07 ; vf09 ; lusso10 ) .
correlations were also found between h@xmath18 fwhm and @xmath0 ( e.g. leighly 1999 ; reeves & turner 2000 ; shemmer et al . 2006 , hereafter : s06 ; shemmer et al . 2008 , hereafter : s08 ; zhou & zhang 2010 , hereafter : zhou10a ) , between @xmath13 and @xmath0 ( e.g. lu & yu 1999 ; porquet et al . 2004 ; wang , watarai & mineshige 2004 ; bian 2005 ; s06,08 ) , and between h@xmath18 fwhm and @xmath13 ( e.g. grupe10 ; paper - i ) .
the correlation between l@xmath21 and l@xmath22 is another important result , which led to further correlation studies related to @xmath7 , l@xmath21 , l@xmath22 and redshift ( e.g. green et al . 2009 , see references given in table [ slope : compare : table ] ) . in table
[ former : study : summary : table ] , we list some of the principal parameters , and list the relevant papers discussing the correlations . later in the papers we will discuss some additional correlations e.g. m@xmath6 vs. @xmath0 and m@xmath6 vs. @xmath14 , in the context that changes in these parameters are caused by changes in the fundamental physical processes .
however , due to the difficulty in obtaining and analyzing both optical / uv and x - ray spectra for a large sample , these parameter correlations are reported separately and based on different samples , rather than being studied systematically for a single well - defined sample .
furthermore , most of the previous samples were not selected based on their spectral properties , so effects such as reddening will introduce biases into the cross - correlations . the lack of a self - consistent physically motivated broadband model , has also been a constraint to performing a more detailed sed study . in paper - i , we defined a sample of 51 type 1 agns with both optical / uv and x - ray spectra which are of high quality , and without evidence of complex spectral absorption features e.g. a warm absorber .
based on this bright and unobscured type 1 agn sample , we applied our latest optical and broadband sed model to perform the spectral fitting , and so matched the optical spectrum and produced a broadband sed for each agn in the sample .
this is so far the most detailed spectral analysis for a medium sized sample of agns , with such well defined high quality spectra . in our preceding paper ( jin , ward & done 2012 , hereafter : paper - ii )
, we studied the balmer emission line properties and the relation between the optical and x - ray emission , based on the same sample and the spectral fitting results given in paper - i .
this paper is based on the sample of paper - i , but focuses on the shape of the broadband sed , and how this relates to a range of parameters from the model .
we will approach this problem by first investigating the numerous correlations previously reported .
then a set of mean seds based on key parameters are constructed and studied in detail .
[ cols="^,^,^,^,^,^,^,^,^,^,^ " , ] | in this third paper in a series of three , we present a detailed study of the broadband spectral energy distribution ( sed ) of active galactic nuclei ( agn ) based on a nearby unobscured type 1 agn sample .
we perform a systematic cross - correlation study of several key parameters , i.e. @xmath0 , @xmath1 , @xmath2 , @xmath3 , @xmath4 , fwhm@xmath5 , m@xmath6 , @xmath7 , @xmath8 and @xmath9 .
the well defined spectral properties of the sample enable us to improve existing relations and to identify new correlations among these parameters .
we confirm a break region around fwhm@xmath5 @xmath10 4000 km s@xmath11 in the @xmath0 vs. fwhm@xmath5 correlation and log(m@xmath6 ) @xmath10 8.0 in the @xmath0 vs. m@xmath6 correlation , where these correlations appear to change form . beyond the break point
the intrinsic @xmath0 index is dispersed around 1.8 .
several new correlations are also reported in this paper e.g. strong correlations in @xmath12 vs. @xmath13 , @xmath12 vs. @xmath14 and @xmath14 vs. m@xmath6 . the principal component analysis ( pca )
is performed on the correlation matrix of the above parameters .
this shows that the three physical parameters , i.e. black hole mass , mass accretion rate and eddington ratio , drive the majority of the correlations .
this is consistent with pca results found from previous optical spectral studies . for each key parameter
, we split the agn into three sub - samples , binned based on increasing value of that parameter .
we co - add the model seds for each object in the sub - sample to see how the sed changes with that parameter .
most parameters , except l@xmath15 , show similar systematic changes in the sed such that the temperature at which the disc peaks is correlated with the ratio of power in the disc versus the comptonised components and the hard x - ray spectral index .
this underlying change in sed shape shows that agn do exhibit intrinsically different spectral states .
this is superficially similar to the sed differences in bhb ( black hole binary ) seen as @xmath13 increases , but the analogy does not hold in detail . only objects with the highest @xmath13 appear to correspond to a bhb spectral state ( the disc dominated high / soft state ) .
the agn with typical mass accretion rates have spectra which do not match well with any state observed in bhb .
we speculate that this could be due to the presence of a powerful uv line driven disc wind , which complicates simple mass scaling between stellar and supermassive black holes .
[ firstpage ] accretion , broadband sed modeling , active - galaxies : nuclei |
the primary goal of this paper is to provide a convenient way of visualising the root systems @xmath0 and @xmath1 .
there are two important relations on a root system that one might wish to have a good understanding of : the poset structure , in which @xmath2 if @xmath3 is a positive root , and the orthogonality structure , in which @xmath4 if @xmath5 and @xmath6 are orthogonal roots . in our paper on cominuscule schubert calculus , with frank sottile @xcite
, we found that our examples required a good simultaneous understanding both these structures .
this is easy enough to acquire for the root systems corresponding to the classical lie groups . in @xmath7 , for example , one can visualise the positive roots as the entries of an strictly upper triangular @xmath8 matrix , where the @xmath9 position represents the root @xmath10
. then @xmath11 , if and only if @xmath5 is weakly right and weakly above @xmath6 .
orthogononality is also straightforward in this picture : @xmath5 and @xmath6 are non - orthogonal if there is some @xmath12 such that crossing out the @xmath13 row and the @xmath13 column succeeds in crossing out both @xmath5 and @xmath6 .
see figure [ fig : staircase ] . in type @xmath14 , it is less obvious how to draw such a concrete picture
. separately the two structures have been well studied in the contexts of minuscule posets @xcite , and strongly regular graphs ( see e.g. @xcite ) . however , once one draws the hasse diagram of the posets , the orthogonality structure suddenly becomes mysterious .
of course , one can always calculate which pairs of roots are orthogonal , but we would prefer a picture which allows us to do it instantly .
thus the main thrust of this paper is to get to figures [ fig : cubecorner ] and [ fig : square ] , which illustrate how one can simultaneously visualise @xmath1 and @xmath0 posets and orthogonality structures , at least restricted to certain strata of the root system .
the restriction of these structures to the strata is exactly what is needed for the type @xmath14 examples in @xcite . with a little more work
, one can use these figures to recover the partial order and orthogonality structures for the complete root system . to reach these diagrams ,
we begin by considering certain maps from a root system to @xmath15 , which are injective ( or 2:1 if @xmath16 ) . once we have some general observations about these maps , we give examples for @xmath0 and @xmath1 which are particularly nice . in these cases , we show that properties of the underlying root system are reflected in simple combinatorial structures on the target space , which is what allows us to produce diagrams in question .
as the @xmath1 example is richer , we will discuss it before the @xmath0 example .
the idea of relating the @xmath0 and @xmath1 root systems to @xmath15 has appeared elsewhere .
for example , harris @xcite uses such an identification to describe the galois group of the 27 lines on the cubic surface
one of the del pezzo surfaces .
the connection between del pezzo surfaces and the exceptional lie groups has been well established ; we refer the reader to @xcite .
one can also see such a relationship reflected in the well known identification of weyl groups ( see e.g. @xcite ) : @xmath17 these facts follow from the identifications outlined in this paper , and presumably have been proved in similar ways before .
let @xmath18 be a simply laced root system , so that with respect to the inner product @xmath19 on @xmath20 we have @xmath21 for all @xmath22 .
we assume that @xmath23 has full rank in @xmath20 .
let @xmath24 denote the lattice in @xmath20 generated by @xmath23 . choose a basis of simple roots @xmath25 , for @xmath26 .
let @xmath27 denote the positive roots with respect to this basis , and @xmath28 denote the negative roots . recall that @xmath27 is a partially ordered set , with @xmath29 iff @xmath30 .
roots @xmath6 and @xmath31 are comparable in the partial ordering iff @xmath32 for each @xmath33 , we define @xmath34 to be the coefficient of @xmath35 , when @xmath6 is expressed in the basis of the simple roots : @xmath36 .
let @xmath37 denote the dynkin diagram dynkin diagram of @xmath23 .
as @xmath23 is simply laced , each component of @xmath37 has type ade .
the vertices of @xmath37 are denoted @xmath38 , and correspond ( respectively ) to the simple roots @xmath39 .
when @xmath23 is a simple root system ( i.e @xmath37 has just one component ) , the affine dynkin diagram @xmath40 is obtained by adding a vertex @xmath41 to @xmath37 , corresponding to the lowest root @xmath42 of @xmath23 .
let @xmath44 be a positive integer
. for reasons explained later in this section , we shall be mostly interested in the case where @xmath45 is a prime , or @xmath46 .
let @xmath47 be a finite rank free module over @xmath43 , with a symmetric bilinear form @xmath48 taking values in @xmath43 .
let @xmath49 suppose that @xmath50 has subset @xmath51 such that @xmath52 then we obtain a map @xmath53 by extending the natural map @xmath54 to a homomorphism of abelian groups .
if @xmath55 then @xmath56 this is true for all pairs of simple roots , and both inner products are bilinear .
suppose @xmath57 .
then @xmath58 @xmath59 @xmath60 . since @xmath61 are roots of a simply laced root system , @xmath62 , thus @xmath58 @xmath59 @xmath63 @xmath59 @xmath64 .
we now restrict the domain of @xmath65 to @xmath23 if @xmath66 and @xmath27 if @xmath16 .
[ thm : injective ] if @xmath67 , the map @xmath68 is injective , and its image lies in @xmath50 . if @xmath69 , the map @xmath70 is injective , and its image lies in @xmath50 .
we first suppose @xmath71 .
note that the fact that @xmath72 is clear from the fact that every @xmath22 satisfies @xmath21 . now , suppose that @xmath73 , @xmath74 .
we show that @xmath75 . for all @xmath76 we have @xmath77 , so @xmath78 .
in particular the set of roots perpendicular to @xmath6 and @xmath31 are equal
. implies @xmath6 and @xmath31 belong to the same simple component of @xmath23 .
there are two cases : if the component is of type @xmath79 , then it is trivial that we must have @xmath75 .
if the component is not of type @xmath79 , then the fact that @xmath6 and @xmath31 have the same set of perpendicular roots implies that @xmath80 .
( in types @xmath81 and @xmath14 , the roots perpendicular to any given root span an entire hyperplane , and in type @xmath82 it is easily checked . )
however , for all @xmath83 , @xmath84 . since @xmath85
, we can not have @xmath86 .
thus @xmath75 . for @xmath16 ,
the fact that every @xmath87 satisfies @xmath21 , implies that @xmath88 .
it is therefore enough to show that @xmath89 is injective .
suppose @xmath90 , @xmath74 .
we show that @xmath75 . as in the @xmath71 case , for all @xmath76
, we have @xmath91 . thus the sets @xmath92 and @xmath93 , where @xmath94 coincide .
this implies @xmath6 and @xmath31 belong to the same simple component of @xmath23 .
( in particular , if @xmath95 then @xmath96 . )
if this component is @xmath97 or @xmath79 , it is trivial that @xmath75 .
if the component is @xmath98 , @xmath99 , then @xmath92 is a root system of type @xmath100 , where @xmath61 are both in the @xmath97 component .
if the component is @xmath0 , @xmath1 , or @xmath101 , then @xmath92 is a root system of type @xmath102 , @xmath103 or @xmath104 , where @xmath61 are both in the @xmath97 component .
thus in both these cases @xmath75 . however , if the component if @xmath105 , @xmath106 , then @xmath92 is a root system of type @xmath107 , where @xmath108 are both in an @xmath97 component ( a priori , not necessarily the same one ) .
if @xmath61 belong to the same @xmath97 component , then @xmath75 .
so suppose they do not .
we identify the roots of @xmath105 with the vectors @xmath109 , where @xmath110 is an orthonormal basis for @xmath111 .
it is easy to see that if @xmath112 , then @xmath113 . so @xmath114 .
on the other hand , for all @xmath115 , we have @xmath116 .
so @xmath117 , for all @xmath118 .
but among these must be a pair of simple roots .
we conclude that @xmath65 restricted to the simple roots is not injective , hence @xmath119 contains fewer than @xmath120 elements , a contradiction .
although we will not have use for it here , if @xmath45 is not a prime , one could also allow the possibility that @xmath47 is not a free module . in this case theorem
[ thm : injective ] remains true provided @xmath121 for all @xmath22 .
this will be the case whenever @xmath122 or when @xmath37 has no component of type @xmath97 .
we now show that the most interesting cases are when @xmath45 is a prime or @xmath46 .
suppose @xmath123 is not prime .
let @xmath124 be a proper divisor of @xmath45 .
let @xmath125 .
let @xmath126 denote the reduction modulo @xmath127 map .
@xmath128 comes with a symmetric bilinear form @xmath129 , the reduction of @xmath48 modulo @xmath127 .
the composite map @xmath130 is injective , and its image lies in @xmath131 .
moreover @xmath132 . as @xmath124
, this follows from the fact that @xmath133 preserves inner products modulo @xmath127 .
the most interesting case of theorem [ thm : injective ] occurs when rank @xmath134 of @xmath47 is smaller than the rank @xmath120 of @xmath26 .
if this is the case , we will call the map @xmath65 a compression of the root system .
here we give a necessary and nearly sufficient condition for compression to be possible .
let @xmath82 be the coxeter matrix of @xmath23 , @xmath135 .
[ prop : pdivides ] if we have @xmath119 as in equation , and @xmath136 , then @xmath45 divides @xmath137 .
let @xmath138 be the @xmath139 matrix whose columns are are @xmath140 in some basis , and let @xmath141 be the @xmath142 matrix representing the bilinear form @xmath48 in the same basis .
then @xmath143 if @xmath144 then @xmath145 , so @xmath146 . conversely , if @xmath45 is prime and @xmath146 , and @xmath147 denotes the reduction of @xmath82 modulo @xmath45 , then one can define @xmath148 .
and @xmath140 is the image of the standard basis vector @xmath149 under the natural map .
this will satisfy , provided the @xmath140 are all distinct and non - zero . the same construction works if @xmath45 is not prime , though @xmath47 will not necessarily be a free @xmath43-module . in particular
, we can not hope for compression in @xmath101 , a root system for which @xmath150 . for @xmath1 , however , @xmath151 , and for @xmath0 , @xmath152 .
thus we should expect compression of the @xmath1 and @xmath0 root systems to be possible , taking @xmath16 or @xmath153 respectively .
the * o - graph * of @xmath47 is the graph whose vertex set is @xmath47 and whose edges are pairs @xmath154 ,
@xmath155 such that @xmath156 .
the * n - graph * of @xmath47 is the complement of the o - graph , having vertex set @xmath47 and edges @xmath154 such that @xmath157 .
as our two main examples involve @xmath16 and @xmath158 , we consider some special inner products @xmath48 in these case .
if @xmath16 , we let @xmath47 be an even dimensional vector space over @xmath159 with a symplectic form @xmath48 . by symplectic form , we mean an ( anti)symmetric non - degenerate bilinear form for which @xmath160 for all @xmath161 .
thus @xmath162 .
we see that @xmath163 satisfies the condition iff the restriction of the n - graph to @xmath119 is isomorphic to @xmath37 . in this case
, the associated map @xmath65 gives an injective map from @xmath27 to @xmath164 .
if @xmath158 , we take @xmath47 to be an @xmath134-dimensional vector space over @xmath165 , with the standard symmetric form @xmath166 note that @xmath167
consider @xmath168 with the standard euclidean inner product @xmath19 .
let @xmath169 be the vectors @xmath170 \alpha_1 & = ( 1,-1,0,0,0,0,0,0 ) \\
\alpha_2 & = ( { { \textstyle \frac{1}{2}}},{{\textstyle \frac{1}{2}}},{{\textstyle \frac{1}{2}}},-{{\textstyle \frac{1}{2}}},-{{\textstyle \frac{1}{2}}},-{{\textstyle \frac{1}{2}}},-{{\textstyle \frac{1}{2}}},-{{\textstyle \frac{1}{2 } } } ) \\ \alpha_3 & = ( 0,1,-1,0,0,0,0,0 ) \\
\alpha_4 & = ( 0,0,1,-1,0,0,0,0 ) \\
\alpha_5 & = ( 0,0,0,1,-1,0,0,0 ) \\
\alpha_6 & = ( 0,0,0,0,1,-1,0,0 ) \\
\alpha_7 & = ( 0,0,0,0,0,1,-1,0 ) \\
\alpha_8 & = ( 0,0,0,0,0,0,1,-1 ) .
\end{aligned } \qquad \qquad \epsfig{file = e8dynkin.eps , height=.6in}\ ] ] these are the simple roots of @xmath101 , which span the @xmath101 lattice .
they correspond to the vertices @xmath171 of @xmath172 , in the order shown on the right .
this ordering of simple roots of @xmath101 corresponds to the inclusion of groups below .
@xmath173 to obtain the root systems of @xmath174 , @xmath175 we take the simple roots to be @xmath39 .
these span the @xmath174 lattice . in general , the roots of @xmath174 are the lattice vectors @xmath176 such that @xmath177 .
the roots @xmath178 of @xmath101 , are stratified as @xmath179 . for @xmath180 , @xmath181 @xmath182
are the roots of @xmath183 minus the roots of @xmath184 .
equation makes sense for @xmath185 , and @xmath186 ; however it is convenient ( and arguably correct ) to put these into the same stratum : @xmath187 we also have a stratification of @xmath188 given by @xmath189 .
this stratification has the property that the roots of @xmath174 , @xmath190 are precisely @xmath191 for each stratum let @xmath194 denote the graph whose vertices are @xmath195 and whose edges form the hasse diagram of the poset structure on @xmath27 , restricted @xmath195 .
thus we have an edge joining @xmath6 and @xmath31 if one of @xmath196 is a simple root .
these are shown in figure [ fig : hasse ] .
let @xmath204 denote the non - cyclic four element group .
we denote the elements of this group @xmath205 , and the operation @xmath206 is binary addition without carry .
@xmath207 is a two dimensional vector space over @xmath159 and thus has a unique symplectic form : @xmath208 we take as our subset @xmath213 , the set @xmath214 , where @xmath215 \begin{alignedat}{2 } s_1 & = 100 , & \quad s_2 & = 030 , \\
s_3 & = 300 , & \quad s_4 & = 111 , \end{alignedat } \\ s_5 = 003 , \quad s_6 = 001 , \quad s_7 = 033 .
\end{gathered } \qquad \qquad \epsfig{file = e7s.eps , height=0.6in}\ ] ] for @xmath225 , the * link * on @xmath226 in the t - graph , denoted @xmath227 , is the restriction of the t - graph to the vertices adjacent to @xmath226 .
the * antilink * on @xmath226 in the t - graph , denoted @xmath228 is the restriction of the t - graph of the vertices non - adjacent to @xmath226 ( not including @xmath226 ) .
note that this result is not independent of the choice of @xmath119 for the images of the simple roots .
we have chosen @xmath119 quite carefully , in part to make this lemma hold .
it is possible , ( and not difficult ) to check this result on each of the 27 roots of @xmath233 ; however , since a symmetry argument is available , we present it here . we know
that @xmath234 , thus it suffices to show that @xmath229 is invariant under the following symmetries : @xmath235\cdot a , b , c ) \\ ( a , b , c ) & \mapsto ( [ 3 \leftrightarrow 2 ] \cdot a , b , c ) \end{aligned}\ ] ] these symmetries come from a dynkin diagram construction , which we first describe for any @xmath174 .
a similar construction can also be used for types @xmath82 and @xmath81 .
let @xmath236 .
we decorate each vertex of @xmath81 with the corresponding simple root in @xmath23 .
choose a vertex @xmath237 , where @xmath238 .
if we delete the edge @xmath239 from @xmath81 , the diagram breaks up into two components @xmath240 where @xmath241 is the component containing @xmath242 .
if @xmath243 , @xmath244 will be empty .
note that @xmath241 is a sub - dynkin diagram of @xmath81 , and hence corresponds to a sub - root system @xmath245 . 1 .
add to @xmath241 the affine vertex @xmath247 , to form the affine dynkin diagram @xmath248 .
this vertex is decorated with the _ lowest root _ @xmath249 .
2 . for every vertex @xmath248 ,
replace the root which decorates the vertex by its negative .
3 . delete the vertex @xmath250 .
if @xmath244 is not empty , reattach it by forming an edge @xmath251 .
the result is @xmath246 .
the underlying graph @xmath246 is isomorphic to @xmath81 , but under this isomorphism , the roots decorating the vertices have changed .
the roots decorating @xmath246 correspond to a new system of simple roots for @xmath23 .
thus this process corresponds gives an automorphism of @xmath23 , and hence to an automorphism of @xmath50 .
returning now to the @xmath1 case , we note that for any root @xmath6 , @xmath252 is preserved modulo @xmath253 under each of these automorphisms .
thus each automorphism restricts to an automorphism of @xmath254 , and hence of @xmath255 .
the symmetries , are the automorphisms of @xmath229 given by the construction above , using vertices @xmath256 respectively .
it is sufficient to check this on the images of the simple roots , and this is easily done .
see figure [ fig : automorphisms ] .
[ thm : t - graph7 ] on @xmath219 , the image of any stratum , the t - graph agrees with the o - graph .
in particular , for @xmath264 , @xmath265 if and only if @xmath266 and @xmath267 agree in exactly one coordinate .
we first show the theorem is true for the stratum @xmath233 . by lemma [ lem:27 ] exactly one of @xmath273 and
exactly one of @xmath274 is zero .
suppose @xmath275 .
then we have @xmath58 @xmath59 @xmath276 and @xmath277 @xmath59 @xmath211 and @xmath221 agree in exactly one coordinate .
suppose @xmath278 . then @xmath58 @xmath59 @xmath279 @xmath59 @xmath211 and @xmath221 agree in exactly one coordinate .
the remaining cases are the same as these two by symmetry . to show the statement for other strata @xmath280 , we consider @xmath281 and @xmath282 . as
@xmath61 belong to the same stratum , we have @xmath283 .
also , @xmath284 is the image of @xmath266 under an automorphism of the t - graph ; thus @xmath285 is adjacent to @xmath286 @xmath59 @xmath266 is adjacent to @xmath267 .
let us calculate @xmath291 .
first note that by theorem [ thm : t - graph7 ] @xmath292 we have @xmath293 @xmath59 @xmath294 @xmath59 @xmath295 and @xmath296 , by .
now since @xmath297 , @xmath298 implies that @xmath295 if and only if @xmath299 .
thus we see that @xmath300 [ lem : orthseq ] let @xmath304 .
if @xmath305 and @xmath306 are orthogonal , there exists a unique vector @xmath307 such that @xmath308 are pairwise orthogonal , and moreover , @xmath309 .
conversely if @xmath310 then @xmath305 and @xmath306 are orthogonal .
in light of theorem [ thm : t - graph7 ] , this is quite easy to show for our preferred choice of @xmath119 .
nevertheless , this result is true for any @xmath119 satisfying , and so we give a more general proof .
let @xmath311 and @xmath312 .
view @xmath313 and @xmath314 as roots in the @xmath101 root system .
assume that @xmath305 and @xmath306 are orthogonal ; hence @xmath315 . throughout the proof we use the fact that the sum of two roots is a root if their inner product is negative . to show existence let @xmath321 denote the affine ( lowest ) root of @xmath101 .
then @xmath322 so @xmath323 is a root .
let @xmath324 and @xmath325 . note that @xmath326 , @xmath327 , so @xmath328 , hence @xmath307 .
and we can explicitly check @xmath329 , so @xmath330 are pairwise orthogonal .
finally , note that the affine root of @xmath101 has the property that @xmath331 for all roots @xmath332 .
thus , for all @xmath332 , we have @xmath333 thus @xmath334 . for uniqueness ,
let @xmath335 be any vector orthogonal to @xmath305 and @xmath306 , and let @xmath336 .
@xmath337 , so @xmath338 is a root of @xmath101 .
but @xmath339 , and the only root of @xmath101 with this property is the affine root @xmath321 .
we conclude that must hold . certainly if one of @xmath351 , then @xmath348 .
suppose that neither @xmath352 nor @xmath353 is in @xmath195 .
note that neither can be in @xmath354 either .
if one of @xmath355 is a root , then it belongs to some stratum other than @xmath195 , so @xmath356 . on the other hand
if neither is a root then @xmath357 .
suppose that @xmath348
. then @xmath358 by corollary [ cor : orthseqcor ] ,
we conclude @xmath359 , a contradiction . one can visualise @xmath229 as the squares one sees looking at the corner of a @xmath360 cube .
the elements are arranged as shown in figure [ fig : cubecorner ] .
we can recover the hasse diagram @xmath361 of the poset structure on @xmath233 in this picture , as follows .
for each simple root @xmath35 , @xmath362 , we draw an edge joining @xmath161 and @xmath363 if @xmath364 . by theorem [ thm : order7 ] we will draw such an edge if and only if the corresponding roots in @xmath229 are related by addition a simple root , which is exactly how the hasse diagram is constructed .
a similar procedure also works on the smaller strata . in this picture , orthogonality is easy to determine as well . by theorem [ thm : t - graph7 ]
this is determined by the links in the t - graph . for any @xmath365
, the set of @xmath366 orthogonal to @xmath161 can be described as follows : if @xmath363 is on the same face of the @xmath360 cube as @xmath161 , then @xmath363 is not in the same row or column as @xmath161 ; if @xmath363 is on a different face from @xmath161 then @xmath363 is in the same extended row / column as @xmath161 .
figure [ fig : cubecorner ] shows @xmath367 , which is set of root images orthogonal to @xmath368 .
suppose @xmath369 and @xmath370 , and @xmath372 .
then @xmath373 if and only if @xmath374 . if @xmath375 then @xmath5 is orthogonal to @xmath6 if and only if @xmath371 is orthogonal to @xmath281 . if @xmath376 then @xmath5 is orthogonal to @xmath6 if and only if @xmath281 is orthogonal to both @xmath371 and @xmath377 , or to neither .
the o - graph restricted to @xmath229 is a well known object ; its complement is the schlfli graph ( see e.g. @xcite for alternate desciptions ) , which describes the incidence relations of the 27 lines on a cubic surface .
it is well known that the full automorphism group of the schlfli graph is the weyl group of @xmath0 .
many of these automorphisms are manifest from our description . if @xmath389 , then the action of the reflection @xmath390 on @xmath229 is given by @xmath391 using theorem [ thm : order7 ] , we see that for @xmath365 , @xmath392 each @xmath390 swaps six pairs @xmath393 and the restriction of the o - graph to these 12 vertices is a union of two @xmath394 graphs , which are maximal cliques .
these pairs are known as schlfli double sixes there are 36 in total , each arising in this way for some unique @xmath395 . from , it is easy to verify that the automorphisms are generated by reflections in the roots @xmath396 ( generating all possible @xmath397 ) ; @xmath398 ( for @xmath399 ) ; @xmath400 ( for @xmath401 ) ; whereas the automorphism corresponds to @xmath402 symmetry of the affine dynkin diagram @xmath403 .
these alone do not generate the weyl group of @xmath0 ; however , together with @xmath404 they do , since this extended list includes all reflections in simple roots .
if @xmath405 is a poset , an * order ideal * in @xmath406 is a subset @xmath407 such that if @xmath408 , and @xmath409 then @xmath410 .
the set of all order ideals in @xmath406 is denoted @xmath411 and is itself a poset , ordered by inclusion .
it is a remarkable fact that the posets @xmath195 are related by such a construction : there is an isomorphism @xmath412 we refer the reader to @xcite for an explanation of this phenomenon . here
we will explore some interesting relationships between this isomorphism and our compression map @xmath65 . 1 .
@xmath416 is a homomorphism of graphs , i.e. if @xmath417 , then @xmath418 ; 2 .
@xmath416 is locally surjective , i.e. for every @xmath419 , @xmath416 maps the neighbours of @xmath226 surjectively to the neighbours of @xmath420 .
it is clear that any isomorphism must be of the form @xmath429 for some function @xmath430 . since the @xmath431 is the minimal element of @xmath195 and @xmath432 and @xmath433 are the two smallest elements of @xmath193 , we must have @xmath434 . in light of proposition
[ prop : uniqueopenmap ] , it suffices to show that @xmath416 must induce an open map from @xmath194 to @xmath425 .
suppose @xmath435 , is an edge of @xmath194 , and let @xmath436 , @xmath437 .
we show that @xmath438 is an edge in the dynkin diagram , i.e. @xmath439 . consider the order ideals @xmath440 , @xmath441 and @xmath442 .
we have @xmath443 , where @xmath444 .
. however , note that @xmath446 is not a root .
if it were then there would two order ideals between @xmath447 and @xmath448 , namely @xmath449 and @xmath450 , which is impossible if @xmath435 .
thus @xmath451 .
we conclude that @xmath452 . in the first case ,
let @xmath463 denote the maximal element of @xmath464 .
note that @xmath465 , and thus @xmath466 is a root .
let @xmath467 .
we have @xmath468 , where @xmath469 . finally , by the definition of @xmath470
, we must have @xmath471 .
figure [ fig : openmap7 ] shows the map @xmath478 pictured on the corner of the @xmath360 cube .
there is a striking symmetry in this picture . in particular
if we impose the equivalence relation @xmath479 and @xmath480 on @xmath481 , the numbers have full @xmath402 symmetry .
this equivalence relation is the one that comes from the involution on the affine dynkin diagram @xmath482 .
furthermore the @xmath402 symmetry is exactly broken by the rule that @xmath483s and @xmath484s are connected to a @xmath485 by a path in @xmath361 on the same face of the cube , whereas @xmath486s and @xmath153s are not .
to understand this symmetry , let @xmath487 , denote the automorphism from the proof of lemma [ lem:27 ] corresponding to @xmath488 , and let @xmath489 denote the automorphism corresponding to @xmath490 .
explicitly , @xmath491 is given by the formula : @xmath492 where @xmath493 is the involution on @xmath482 , taking @xmath494 , @xmath495 , @xmath496 , and fixing @xmath497 , @xmath498 . under the identification of @xmath233 with @xmath229 , @xmath491 and @xmath133 are more simply described by @xmath499 , and @xmath500 .
the involution @xmath491 induces an involution @xmath501 on order ideals : @xmath502 note that the root @xmath503 , which is mutually orthogonal to @xmath504 and @xmath505 , is a fixed point of @xmath491 .
we use @xmath506 to partition @xmath507 into four disjoint sets : @xmath508 in fact @xmath501 gives a bijection between @xmath509 and @xmath510 . note that @xmath511 can be identified with the order ideals in @xmath512 , whereas @xmath513 can be identified with the order ideals in @xmath514 .
each order ideal in the latter is a @xmath515 rotation of an order ideal in the former .
thus we also have a bijection @xmath516 defined by @xmath517 since @xmath518 , we must have corresponding structures on @xmath519 , which are related by @xmath520 .
indeed we have an involution @xmath521 on this set , defined by @xmath522 where @xmath523 . the image of each @xmath509 is simply described as @xmath524 and we have a map @xmath525 given by @xmath526 where @xmath527 . to see the near rotational symmetry , we consider @xmath535 , with @xmath536 .
then @xmath537 .
we see that @xmath531 for some @xmath12 , whereas @xmath538
. thus @xmath539 and @xmath540 will be reflections of each other in @xmath482 .
the discussion in sections [ sec : strata ] and [ sec : order ] gives a description of the strata @xmath280 for all @xmath541 , but it is not the most symmetrical one for @xmath290 . for @xmath542 , it is easy to obtain nice description of the strata , as they are subsets of the @xmath543 root system . for @xmath544 , we can obtain a pleasant description by working over @xmath165 .
let @xmath23 be the @xmath0 root system .
let @xmath545 , with the standard symmetric form .
let @xmath546 , where @xmath547{2 } s_1 & = 12000 & \quad s_2 & = 00012 \\ s_3 & = 01200 & \quad s_4 & = 00120 \\ s_5 & = 00011 & \quad s_6 & = 11111 \end{alignedat } \qquad \qquad \epsfig{file = e6s.eps , height=0.6in}\ ] ] it is an injection by theorem [ thm : injective ] . to show it is a bijection
, we must calculate the size of @xmath50 .
the vectors in @xmath50 have either @xmath253 or @xmath483 non - zero coordinates , each of which can be @xmath549 .
thus there are @xmath550 elements in @xmath50 .
but @xmath551 , so @xmath65 is a bijection .
the argument is parallel to the proof of lemma [ lem:27 ] .
one can check that the automorphisms corresponding to dynkin diagram vertices @xmath556 give permutations of the coordinates which generate the symmetric group @xmath557 , hence @xmath557 acts on the coordinates of @xmath558 .
furthermore , the automorphism corresponding to @xmath559 is @xmath560 . applying these automorphisms to @xmath561
, we see that all 16 elements of @xmath558 are indeed of the correct form . certainly if @xmath349 , then @xmath567 .
suppose that @xmath568 .
if @xmath569 is a root , then it belongs to some stratum other than @xmath195 , so @xmath570 . if @xmath571 , then @xmath572 is the image of a negative root , so @xmath570 .
otherwise , since @xmath352 is not a root , we must have @xmath573 . in this case ,
@xmath574 , so in fact @xmath575 .
as we did with @xmath233 , we can recover the partial order structure on @xmath576 ( and the smaller strata ) using theorem [ thm : order6 ] .
if the elements of @xmath558 are arranged as shown in figure [ fig : square ] , we join @xmath161 and @xmath363 if @xmath577 is a simple root . in light of theorem [ thm : t - graph6 ] , orthogonality is also easily determined in this picture .
figure [ fig : square ] shows the example of @xmath578 , which gives the set of root images orthogonal to @xmath579 . | we examine certain maps from root systems to vector spaces over finite fields . by choosing appropriate bases
, the images of these maps can turn out to have nice combinatorial properties , which reflect the structure of the underlying root system .
the main examples are @xmath0 and @xmath1 . |
the principal aim of this work is to study the evolution of morphology in hydrodynamical simulations of galaxy clusters , for flat cdm cosmological models , normalized in order to reproduce the observed cluster abundance .
we also compare its trend with observational x ray data and find a substantial discrepancy on the rate of evolution of substructures , which seem to evolve faster in models than in data .
this result is not surprising , as the slope of cdm spectrum , around the cluster scale , is known to be too flat .
e.g. , the _ extra power _
parameter @xmath5 ( @xmath6 are mass variances on @xmath7mpc scales ; @xmath8 ) is significantly lower in data than in model predictions : for apm galaxies , peacock and dodds ( 1995 ) found @xmath9 ; for the abell / aco sample , borgani ( 1997 ) found @xmath10 in the interval 0.180.25 ; on the contrary cdm models predict @xmath11 .
what is however stricking is the sensitivity that the tests we performed on hydrodynamical simulations , seem to have in appreciating the spectral slope .
high @xmath10 values imply that too many intervene in the dynamical growth of fluctuations , on cluster scales
. an excess evolution can be therefore expected .
but available data cover only a narrow redshift interval ( @xmath12 ) and simulation outputs shall be used on the same interval , reproducing data distribution .
the point is then that different rates seem already visible on such redshifts , which concern a fairly recent cosmic evolution .
the tests we use start from the a statistical tool introduced by buote & tsai ( 1995 ) , who defined the so called power ratios , which essentially derive from a multipole expansion of x ray surface brightness . in sec .
2 more details on definition will be reported .
much work on was already done , both in order to derive them from data and to compare them with simulations ( buote & tsai 1995 , buote & tsai 1996 , tsai & buote 1996 , buote & xu 1997 ) .
in a recent work , buote e xu ( 1997 ) gave some plots where the evolution of cluster morphology in data and cdm simulations are compared ; similar plots are given also by us . here
, however , we present two essential improvements : ( i ) instead of using pure cdm n body simulations , we use the outputs of a treesph code , therefore including hydrodynamical effects .
a large number of massive clusters are treated in this way and hydrodynamical effects are found to be important in shaping the distribution of the x ray emitting baryon component .
this will be shown in some plots where dm and baryon distributions are compared ; but there is also a back effect of the different baryon distributions on the evolution of the potential well , which can not be appreciated directly from the plots .
( ii ) furthermore , to compare data and simulations , we use an advanced statistical discriminator ( peacock 1982 , fasano & franceschini 1987 ) , which allows us to find quantitative estimates of the probability that the same process gives rise to observational and simulated statistical outputs .
the discrepancy we find is however directly visible in suitable evolutionary plots and can be also appreciated using a simpler statistical test .
the plan of the paper is as follows . in sec .
2 power ratios are defined and their evolution is discussed . in sec .
3 we give some details about the model and simulation procedure followed to define clusters and study their evolution .
power ratios are computed in sec .
4 where we also perform the comparison between data and simulations .
results are discussed in sec . 5 , where some general conclusions are also drawn .
appendix a includes a brief outline of the pf2 statistical test , and some discussion of the reasons why its extension to three dimensions is premature .
in this section we shall report results of tsai & buote ( 1996 ) and buote & xu ( 1997 ) , which are the starting point of our arguments .
the steps to work out the power ratios from a model cluster are the following ones : ( i ) we assume that the x ray emission power is @xmath13 and project it on a random plane @xmath14 , yielding a surface brightness @xmath15 ( apart of a proportionality factor ) .
( ii ) we find the projected mass center ( or _ centroid _ ) @xmath16 for @xmath15 ; in respect to @xmath16 , the surface brightness can be expressed as @xmath17 ; ( iii ) we use such @xmath18 to generate a pseudo
potential @xmath19 , by solving the poisson equation : @xmath20 ( again , a constant factor , in front of the r.h.s .
@xmath21 , the gravitational constant @xmath22 , would turn out to be irrelevant ) .
( iv ) we expand @xmath23 in multipoles ; from the coefficients of such expansion , the power ratios @xmath24 are obtained .
more in detail : @xmath25 where @xmath26 ^ 2\ ] ] and @xmath27 \cos(m\varphi ) } } \ ] ] @xmath28 \sin(m\varphi)}}\ ] ] henceforth , the constants which were unspecified cancel out , when the ratio of @xmath29 s is considered .
moreover , owing to the definition of the centroid @xmath16 , @xmath30 vanishes and @xmath31 should then be disregarded . as we restricted the analysis to @xmath32 , which account for on scales not much smaller than @xmath33 itself
, there are only 3 significant ( @xmath34 )
. therefore , during its evolution , a cluster moves along a line ( _ evolutionary track _ ) in the 3dimensional space spanned by such s , whose 2dimensional projections will be considered below . starting from a configuration away from the origin , corresponding to a large amount of internal structure
, a cluster evolves towards isotropization and homogeneization .
this motion , however , does not occur with a steady trend , as sudden bursts of structure appear when further matter lumps approach the cluster potential well , to be absorbed by it .
however , the evolutionary track eventually approaches the origin , and this can be more easily appreciated by averaging over the contributions of several clusters .
actual data , of course , do not show the motion of a single cluster along the evolutionary track .
different clusters , however , lie at different redshifts and , in average , can be expected to describe a succession of evolutionary moments .
power ratios for model clusters are to be set in the 3dimensional space spanned by s , taking each model cluster at redshifts distributed as for data clusters .
the data set is the same of buote @xmath35 tsai ( 1996 ) . among x
ray cluster images , taken with @xmath36 pspc ( pfeffermann 1987 ) , and such that the pspc central ring contains a cluster portion whose radius exceeds 400@xmath37kpc , they took those contained in the heasarc legacy database and belonging to the ebeling ( 1993 ) or edge ( 1990 ) samples . out of the 59 objects selected in this way , they could estimate for 44 of them at @xmath38mpc and for 27 of them at @xmath39mpc .
pspc data give the x ray surface brightness @xmath40 , which is to be used in the same way as the @xmath18 s obtained from models , to work out for @xmath41mpc . in fig .
1 we give the redshift distribution of data clusters , for the three values of @xmath33 . as we shall discuss further in the next section , we have 40 simulated clusters to deal with . we take each of them , only at a single @xmath42 value , producing 50 @xmath42 distributions similar to real data . in order to work out power ratios , we need to treat each cluster with reference to a given plane : 50 random planes are selected for each cluster and one of them is considered for each @xmath42 sequence .
the main contribution of this work is given by the simulations we run , to work out , and that will be however used for several other aims .
we started from a box with a side @xmath43mpc , where we evolved a cdm model with @xmath44 and @xmath45 .
the initial spectrum was normalized in order that the cumulative cluster number density @xmath46 , for @xmath47 , yields @xmath48 , according to data on cluster abundance ( white 1993 , biviano 1993 , eke 1996 ) .
this corresponds to @xmath49 .
the simulation was run from @xmath50 to @xmath51 using a p3 m code with 10@xmath52 particles ( mass @xmath53 ) .
the choice of such @xmath54 value is dictated by the opportunity of restarting from the same conditions when hydrodynamic simulations are then performed .
the pm part of the code makes use of 256@xmath55 cells and a smoothing radius of 520 kpc was taken for gravitational forces . at @xmath51 clusters were then identified using a fof algorithm ; we considered as _ friends _ , particles at a distance @xmath56 ( @xmath57 is the mean particle separation ) .
the 40 most massive clusters were then located ; the lightest one is made by 34 particles and has a mass @xmath58 .
for each of these clusters an hydrodynamic simulation was then performed , using a treesph code .
the hydrodynamical part of the code is based on the sph method ( hernquist @xmath35 katz 1989 ) .
the gravitational forces are solved with a hierarchical tree algorithm and this part of the code is based on the public treecode of l. hernquist ( hernquist 1987 ) . in order to run the hdyrodynamical simulation we located the cluster center at @xmath51 and found all particles within a radius @xmath59 , where the cluster density is @xmath60 ( background density ) .
such particles were traced back to @xmath54 , in the original simulation cube , and a smaller cube of side @xmath61 , enclosing all of them , was then located at the cluster center .
we found the approximate scaling @xmath62 .
a high resolution lattice of @xmath63 particles was set in the smaller cube and their position was then perturbed , using the same initial conditions of the cosmological simulations , implemented by additional waves to sample the increased nyquist frequency .
this technique is similar to the one adopted by katz & white ( 1993 ) , navarro , frenk & white ( 1995 ) .
we used two grids , for baryon and cdm particles , whose masses @xmath64 .
the cube of side @xmath61 is then set inside a greater cube of side @xmath65 , with parallel sides and equal center .
such greater cube is initially filled by @xmath63 cdm particles , whose spacing is double in respect of inner cube particles , and whose mass is @xmath66
. then particles falling in the smaller inner cube are eliminated .
heavier particle positions are also perturbed , according to the initial conditions of the cosmological simulation .
finally , we use all particles inside a sphere , of radius @xmath61 and with the same center as cubes , for the high resolution treesph simulation .
since we take a constant number of grid particles , their masses vary according to @xmath61 ; for the most massive cluster we met ( @xmath67 ) , it was @xmath68 .
the softening parameter for gravitational forces ( @xmath69 ) was set to @xmath70 for gas particles and scales as @xmath71 .
we use a tolerance parameter @xmath72 , without quadrupole corrections .
viscosity was set as in hernquist @xmath35 katz ( 1989 ) taking their parameters @xmath73 and @xmath74 .
the minimum timestep was @xmath75 for gas particles and twice as much for cdm .
in average each cluster required @xmath76hours of cpu time to be evolved from @xmath54 to @xmath51 on a risk 6000 elaborator .
outputs from treesph simulations were preserved at various redshifts . among the smaller ones
, we used @xmath770.15 , 0.10 , 0.049 and 0 ( @xmath78 ) . in fig .
2 we give a typical example of projected @xmath79 outputs , at a set of redshifts @xmath80 . already from such figure , it is clear that cdm clusters are rich of substructures and characterized by a fast substructure evolution . in fig .
3 we show cdm and baryon distributions at @xmath51 for a typical cluster .
such figure shows that the baryon distribution is significantly smoothed out in respect to cdm .
this obviously reflect on s , although we shall report no quantitative details on this point here .
clusters are expected to evolve by accreting lumps of baryons and cdm onto a deep potential well . as time elapses ,
nearby lumps are gradually absorbed and cluster substructures tend to vanish .
such behaviour is clearly visible also in fig . 2 .
in fig . 4 , instead , we give 9 plots for the space containing the evolutionary tracks projected on 3 planes and at three different @xmath33 s .
cross indicate the location of data clusters .
filled dots refer to a single selection of simulated clusters .
the trend visible in this selection of simulated clusters is however a general feature : a linear regression gives place to the dotted straight line reported in the plots .
although real data do not show such a marked trend , a linear regression was performed also on them .
the continuous lines show the results of such linear fits to data .
.angular coefficients and standard deviations for linear fits .
@xmath82 refers to data and simulations . in parenthesis
we give also simulation errors from 20 bootstrap resamplings . [ cols="^,^,^,^,^,^",options="header " , ] the extra input that we should eliminate , to obtain the probability that the observational data set is fit by artificial data sets , is the requirement of linear behaviour .
this can be done by applying the generalization of the kolmogorov
smirnov test to 2dimensional data sets , proposed by peacock ( 1982 ) and widely tested by fasano @xmath35 franceschini ( 1989 ) .
we shall refer to this estimate as pf2 test and more details on it will be given in appendix a. in table 3 we give the values of the single case of fig .
4 and the averaged values of the pf2probabilities that the observational data set and each artificial data set are realizations of the same process , for each @xmath83 pair and each @xmath33 value .
such average pf2probabilities are not so small as those reported in table 2 and essentially correspond to a @xmath84 discrepancy . however , even this sophisticated test is unable to work out a final value which takes into account all the above partial results .
there can be however little doubt left that such a sequence of values , where a faster evolution in model in respect to data is systematic , could hardly be due to observational sample biases or to disregarding peculiar physical effects in the simulations , and is more likely to indicate that the model is not adept to reproduce real data .
it has been known for several years that cobe normalized cdm yields a cluster abundance exceeding observation by a factor @xmath85 . in order to obtain a fair number of clusters , the quadrupole data has to be lowered down to @xmath86k , at more than 3@xmath87 s from the observational value .
however , cdm models are still currently in use as reference models ; in fact , 0cdm models , mixed models or @xmath88 models , used to obtain fair fits of observational data on various scales , involve more parameters . according to our results , however , even though cdm is normalized to yield a fair number of clusters , we can not expect their morphologies to be adequately approached by the model .
this output is related to the slope of the transfered spectrum , as already outlined in sec . 1 , and as is expressed by the values of the _ extra power _
parameter @xmath10 .
it is however important and fairly unexpected that the effects of such different slope are already visible in the average evolutionary tracks , for a fairly narrow redshift range ( @xmath89 ) as the one we had in observational data . in a forthcoming paper , based on work in progress
, we shall extend this analysis to other cosmological models , whose transfered spectrum has a @xmath10value consistent with observations .
this is important , in order to test how sensitive the power ratios @xmath83 are to the @xmath10 value and to check whether this test depends only on @xmath10 or , @xmath21 , also on the @xmath90 of the model considered .
it is a pleasure to thank david buote for his useful comments .
one of us ( s.g . ) wishes to thank sissa for its hospitality during the preparation of this work .
0.3truecm as is known , the kolmogorov smirnov ( ks ) test is used to evaluate the probability that 2 distributions on a single variable @xmath91 are produced by the same process . if @xmath92 events are located at @xmath93 ( @xmath94 ) , let @xmath95 give the fraction of events located at @xmath96 .
clearly @xmath97 is discontinuous in each @xmath93 , where it increases by @xmath98 .
let two sets of @xmath99 and @xmath100 events yield @xmath101 and @xmath102 ; the ks statistics is then defined as the maximum distance between them : @xmath103 and it can be shown that the probability that the two data sets are produced by the same process can be estimated , by using @xmath104 , as follows : let be @xmath105 and @xmath106 .
then such probability reads : @xmath107 ~. \eqno ( a3)\ ] ] various numerical routines exist which allow to evaluate such function with any desired approximation ( see , @xmath21 , press 1992 ) .
while for one dimensional distributions the meaning and the building of @xmath104 are simple ( it tells us the maximum difference of integrated distributions ) , in the case of two dimensional distributions the events are not ordered and there is no immediate way of telling where such difference is maximum .
a further problem is set by the degree of correlation amongst data .
a limiting situation occurs if they locate along a straight line on a plane , and then the ks test would be suitable again .
the problem was considered in detail by peacock ( 1982 ) and fasano @xmath35 franceschini ( 1987 ) , whose approach ( pf2 ) we shall briefly report here .
if @xmath92 data are located on the 2dimensional plane @xmath108 and have average @xmath109,@xmath110 , let us first define @xmath111 and @xmath112 .
their correlation coefficient is then : @xmath113 ( notice that , for @xmath114 , the points would actually lie on a straight line ) .
then , if we have two sets of @xmath99 and @xmath100 data , with correlation coefficients @xmath115 and @xmath116 , defined according to eq .
( a4 ) , we define @xmath117 as square average correlation . the basic point , however , amounts to define a 4value extension of the function @xmath118 , for each set of @xmath92 data : the 4 values are the fractions of data in each of the four _ quadrants _ set by the point @xmath108 where @xmath118 is evaluated . the pf2 statistics @xmath104 will be the maximum difference between @xmath101 and @xmath102 values , ranging both over points and quadrants . using such @xmath104 , the square average correlation @xmath119 , and @xmath120 , defined according to eq .
( a2 ) , we then define @xmath121 \eqno ( a6)\ ] ] and , using eq .
( a3 ) , we obtain the required probability .
it ought to be outlined that there are some restrictions to the significance of these probability estimators .
in general , they become safer as @xmath120 increases . however , in the one dimensional case , the estimate is good for @xmath12220 ( corresponding to @xmath123 ) . in the two dimensional case ,
the restriction is already much stronger , as it must be @xmath124400 ( corresponding to @xmath125 ) , and the further condition that the probability found is @xmath126 is to be added .
our sets of data and simulation points are therefore just adequate to meet the above conditions .
this is important also in view of a step forward that could be considered : treating directly the 3dimensional distributions in the parameter space spanned by @xmath127 ( @xmath34 ) , instead of their two dimensional projections .
an algorithm extending the above test to 3dimensions is a straightforward generalization ; the range of validity of such algorithm , instead , should be accurately tested
. a large portion of the work of fasano and franceschini ( 1987 ) amounted to perform such checks , for the two dimensional case .
however , the trend found beween 1 and 2 dimensions , seems to indicate that , in three dimensions , we need 2 sets of data containing @xmath128 points . as far as simulations are concerned , this requires just doubling our simulation set .
data clusters , instead , are signicantly below such limit . | we compare the evolution of morphology in x ray cluster data and in clusters obtained in a simulation of flat cdm , normalized to the observed cluster abundance .
we find that the evolution rate in model clusters is significantly higher than in data clusters .
the results is stricking , as our cluster data are all just for @xmath0 , but is not contrary to expectations , as is known that the cdm spectrum is too flat , around the cluster scale , and therefore induces the presence of too many substructures .
the cluster simulations were run using a treesph code and therefore include hydrodynamical effects . the test we performed , which turns out to be very sensitive , starts from the so called power ratios introduced by buote and tsai , and makes use of the 2dimensional generalization of the kolmogorov
smirnov test .
the discrepancy between data and model is however directly visible through linear regression techniques and using simpler statistical tests .
2@xmath1 3@xmath2 4@xmath3
80@xmath4 galaxies : clusters : general galaxies : evolution x ray : galaxies cosmology : theory cosmology : simulations . |
the discovery of the extrasolar planets is opening a new world beyond our solar system .
ever since 1995 , mayor & queloz ( 1995 ) detected the first extrasolar giant jupiter51 peg , and to date there are more than 100 planetary systems discovered(marcy , cochran , & mayor 2000 ; butler et al .
2003 ) using the radial velocity technique in the surveys of nearby young stars . at the time of writing ,
a dozen of multiple planet systems
hd 82943 , gj 876 , hd 168443 , hd 74156 , 47 uma , hd 37124 , hd 38529 , hd 12661 , 55 cnc , @xmath19 and ( fischer et al . 2003 ) , hd 169830 ( mayor et al . 2004 ) and hd 160691 ( jones et al .
2002 ) were discovered in recent years and this number is undoubtedly cumulative as the measurements are being carried on .
hence , it is necessary to categorize the discovered multiple planetary systems according to their statistical characteristics ( such as the distribution of the planetary masses , semi - major axes , eccentricities and metallicity ) ( marcy et al .
2003 ) , then to study the correlation between mass ratio and period ratio ( mazeh & zucker 2003 ) and further to improve the understanding of the relationship between the planet occurrence rate and stellar metallicity ( santos et al . 2003
; fischer , valenti , & marcy 2004 ) .
the other key point is to investigate possible stable configurations for the multiple systems , in which the observations reveal that most of them are typically characterized by mean motion resonance ( mmr ) and ( or ) apsidal phase - locking between their orbiting companions ( fischer et al .
2003 ; ji et al . 2003a
; lee & peale 2003 ) , so that one can better understand the full dynamics of these systems . in the present study
, we mainly focus our attention on the hd 82943 system . as of april 4 , 2001 , the geneva extrasolar planet search teamudry / planet / hd82943syst.html ] announced the discovery of the hd 82943 planetary system at an eso press release , consisting of two jupiter - like planets orbiting the parent star .
subsequently , israelian et al . ( 2001 , 2003 ) reported the discovery of @xmath20li in the atmosphere of this metal - rich solar - type star and indicated that the presence of @xmath20li can probably be shown as evidence for a planet or planets having been engulfed by the star hd82943 . at first , let us review several stellar features for this star : hd 82943 is a g0 star with @xmath21 , hipparcos parallax of 36.42 mas , distance of 27.46 pc , and [ fe / h ] = @xmath22 ( santos et al .
2001 , 2003 ) .
the star is aged 2.9 gyr with the stellar mass of 1.15 @xmath23 .
similar to the gj 876 system ( marcy et al .
2001 ) , which is believed that stability of the planetary system might be sustained by the 2:1 mean - motion resonance ( mmr ) and apsidal alignment in the semi - major axes ( kinoshita & nakai 2001 ; lee & peale 2002 ; ji , li , & liu 2002 ) , the two planets of hd 82943 are now also close to a 2:1 commensurability , with orbital periods @xmath24 and @xmath25 d , and semi - major axes 1.18 and 0.75au ( mayor et al . 2004 ) .
however , the previous investigations ( gozdziewski & maciejewski 2001 ) of the stability of the hd 82943 system showed that the system given by the earlier best - fit orbital solution was extremely unstable due to the strong interaction of two massive planets with high eccentricities , hence a natural question is that whether the two companions for this system are really in a 2:1 orbital resonance and whether the system is secularly stable with the observed best - fit orbital parameters . additionally , the stability for hd 82943 places constraints on both planetary masses if the mass of the outer companion is smaller than that of the inner planet ( hadjidemetriou 2002 ) , a system that harbors two planets moving on elliptic 2:1 resonant periodic orbits will be destabilized as a finale . nevertheless , the recently updated orbital solutions ( mayor et al .
2004 ) , with more abundant observations as of october 13 , 2003 , show that the inner planet is a bit more massive than the outer one , thus it is indeed noteworthy to thoroughly reanalyze the observed hd 82943 system , still these mentioned issues of this fascinating system are to arouse our great interests and inspire us to seek after the answers .
consequently , our first goal is to carefully explore the system to fully understand in what kind of likely configurations that two companions may orbit about their host star and if exist , to reveal the dynamical mechanisms to remain the system . in a viewpoint of celestial mechanics ,
the former studies are concerned for the dynamical analysis for the global stability of the exosystems ( gozdziewski & maciejewski 2001 , 2003 ; ji et al .
2002 ; kiseleva - eggleton et al .
2002 ; barnes & quinn 2004 ) , the investigations of resonant picture in the phase space ( hadjidemetriou 2002 ; ji et al . 2003a
; haghighipour et al . 2003 ; callegari jr . , michtchenko , & ferraz - mello 2004 ) or chaotic behavior ( jiang & yeh 2004 ) in these systems , and the existence of habitable zones ( hz ) for earth - like planets that supply stable liquid - water environment to cultivate extraterrestrial intelligent beings ( kasting , whitmore , & reynolds 1993 ; jones & sleep 2002 ; laughlin , chambers , & fischer 2002 ; dvorak et al . 2003 ; menou & tabachnik 2003 ) .
other recent works on the 2:1 resonant geometry of the extrasolar planetary systems include those by hadjidemetriou & psychoyos ( 2003 ) , beauge , ferraz - mello , & michtchenko ( 2003 ) and lee ( 2004 ) .
they studied the presence and location of stable equilibrium solutions or dynamical evolution of such exosystems in more general way . in this work
, we revisit the dynamics of the hd 82943 system and outline several possible stable configurations for a system harboring two companions in a 2:1 resonance , and our new findings are that this system can be locked in aligned or antialigned orbital motions , which these results are never presented nor analyzed in other studies .
specifically , at first , we introduce the numerical setup ( see 2 ) for the dynamical simulations . in 3 , we make a study of the various planetary configurations for hd 82943 by means of the semi - analytical treatment and direct integrations over long - term orbital evolution .
we present the numerical results of hd 82943 by employing two series of the best - fit orbital solutions , and we discover three kinds of stable planetary geometry for this system that all are linked to a 2:1 resonance . furthermore , we advance a semi - analytical model that can avoid the difficulties in the perturbation expansions for the larger eccentricities and still help to explain the numerical evolutions . still , we investigate how the stability for the resonant topology depends on the planetary mass ratio and the orbital parameters on the basis of the new fit ( mayor et al .
2004 ) . in 4 ,
we explore whether there exist an earth - like planet surviving about the habitable zones for the authentic systems of hd 82943 and gj 876 . the other extraordinary phenomenon in the exosystems
is that many of them can host giant jupiter - like planets with strikingly larger eccentricities , and in figure 1 is shown the distribution of the eccentricities of 111 planets discovered at present day .
the figure exhibits that more than 50% of the planets have the eccentricities larger than 0.30 , and hd 80606 b ( naef et al . 2001 ) can occupy the eccentricity amounting up to 0.93 , which indicates that such unusual discoveries are quite different from the circumstances in our solar system , where most of the major planets revolve around sun on the near - circular orbits .
in addition , from the view of the observations , the larger eccentricities seem to favor the planet detectability , where the semi - amplitude of wobble velocity @xmath26 ( with @xmath27 ) , herein @xmath28 , @xmath29 , @xmath30 , @xmath31 and @xmath32 are , respectively , the stellar mass , the planetary mass , the orbital semi - major axis , the eccentricity and the inclination of the orbit relative to the sky plane , requires more massive planetary mass ( @xmath33 the order of jupiter s mass ) , a lower @xmath30 ( e.g. , close - in giant planets ) and a higher @xmath31 ( see fig .
moreover , many previous authors have put forward diverse theories and possible mechanisms to interpret the evolution and origin of the orbital eccentricity variation in the exosystems : the disk - planet or planet - planet interaction ( goldreich & tremaine 1980 ; weidenschilling & marzari 1996 ; lin & ida 1997 ; ford , rasio , & sills 1999 ) render the orbital migration of gas giant planets embedded in the protoplanetary systems ( lin , bodenheimer , & richardson 1996 ; ward 1997 ; bryden et al .
2000 ; nelson et al .
2000 ) and the eccentricities are usually believed to excite to the observed values through complicated processes ( snellgrove , papaloizou , & nelson 2001 ; kley 2003 ) or hybrid mechanisms ( lee & peale 2002 ; chiang , fischer , & thommes 2002 ) related to both the planetary migration and the resonant capture . therefore , as a second objective , we intend to preliminarily make a discussion of the possible origin of the eccentricities or resonant configurations ( see 5 ) for two planets of hd 82943 .
as a final part , in 6 , we summarize our principal results and give a brief discussion .
in the present work , we aim to numerically investigate the orbital motions for the two companions for hd 82943 in a three - dimensional space . and
at first , let us bear in mind that the two companions of this system are assumed to be in the same orbital plane for our simulations except where noted .
here we adopt the best - fit orbital parameters from the web site of the geneva team .
as is known , the two - kepler fit produce five parameters set-(@xmath34 , @xmath35 , @xmath36 , @xmath37 , @xmath38 ) , where for each planet @xmath39 , are the amplitude @xmath34 , the orbital period @xmath35 , the eccentricity @xmath36 , the argument of periapse @xmath37 and the time of periapse passage @xmath38 .
the geneva team presented two sets of the orbital data and hereafter we respectively call them fit 1 and fit 2 . here for fit 1 of 107 observations ( as of july 31 , 2002 ) with the residual of 7.4 m / s , the data are listed here : @xmath40 @xmath41 , @xmath42 @xmath41 , @xmath43 au , @xmath44 au , @xmath45 , @xmath46 , @xmath47 and @xmath48 , where hereafter the subscripts 1 and 2 denote the inner and outer planets , respectively . and fit 2 were derived by fitting 142 observations ( as of october 13 , 2003 , see also mayor et al . 2004 ) with the residual of 6.8 m / s ,
the orbital parameters are presented : @xmath49 @xmath41 , @xmath50 @xmath41 , @xmath51 au , @xmath52 au , @xmath53 , @xmath54 , @xmath55 and @xmath56 . apparently , the eccentricities , semi - major axes , periapse arguments , true anomalies and others can act as key factors to determine the shape of the radial velocity curves .
however , due to the large interaction between two planets in a low mean motion resonance , laughlin & chambers ( 2001 ) pointed out that short - term perturbations among massive planets in multiple planet systems ( such as gj 876 , hd 82943 ) can result in radial velocity variations of the central star that differ substantially from velocity variations derived assuming the planets are executing independent keplerian motions . on the other hand , the best - fit orbital solutions can vary a little from time to time provided that more updated observations were supplied or other elaborate fitting procedure was considered ( mayor et al .
2004 ) , or the velocity trend is removed due to the unveiling of an additional planet or more planets ( fischer et al .
2001 ) for a specific system .
therefore , at this stage , it is meaningful to study the dynamical characteristics of the system based on the subsequent points : ( i ) slightly varying the orbital parameters about the best - fit parameters by adding the uncertainties with direct integrations , ( ii ) seeking for the possible stable geometry for hd 82943 in parameter space , ( iii ) the dynamical mechanisms of retaining the system and ( iv ) the likely eccentricity or configuration origin for two planets , which motivate us to dedicate to this contribution .
next , we continue to describe the means of preparation for the initial orbital data before we start the numerical integrations : for each planet , we usually need to obtain six orbital elements - the semi - major axis @xmath30 , the eccentricity @xmath31 , the orbital inclination @xmath57 , the nodal longitude @xmath58 , the argument of periastron @xmath59 and the mean anomaly @xmath60 . here , two groups of the initial orbital elements were produced on the basis of the data by fit 1 and fit 2 . in each group , we assume that the semi - major axes of the two planets are always unchanged , say , for fit 2 , being 0.75 and 1.18au respectively . as we have indicated above
, the two resonant planets are supposed to be coplanar for the initial configuration and the inclinations are both taken as small constant values not far from zero , e.g. , @xmath61 .
the eccentricities and arguments of periapse are generated in the orbital parameters space in the proximity of the best - fit orbital solutions given the nominal observation errors .
for example , for fit 2 , we take the observed eccentricities @xmath62 or @xmath63 to be centered , respectively , at 0.38 and 0.18 , and randomly displaced by not more than @xmath64 , respectively , then we obtain the resulting initial eccentricities for integration .
still , we carried out similar steps to achieve the starting values of periapse arguments .
the remained two angles of nodal longitudes and mean anomalies , @xmath65 and @xmath66 , such that the linear combination of initial arguments can be close to the resonant configuration , this is reasonable , because the real system is observably stable and approximate to 2:1 mmr . and in table 1 and 2 , we present the orbital solutions where the measurement errors of @xmath67 and @xmath68 , indicating that the eccentricities and periapse arguments are just equal to those by the geneva team . ] are randomly distributed between @xmath69 and @xmath70 . as a consequence
, we obtained 100 orbits for each planet , to perform the integration of the planetary system , and each pair of the orbits was integrated for the time span of @xmath71 yr , unless otherwise stated .
obviously , these orbits just lie in the neighborhood of the fitting parameters and represent the orbital motions quite close to the reality . here
we make study of the possible motion near the best - fit orbital solutions and expect that such investigations can be helpful to reveal some important dynamical features for the studied system or present valuable clues on the searching for other exosystems related to a 2:1 resonance . with the n - body codes ( ji et al .
2002 ) , we carried out the numerical integrations for the hd 82943 system . in our simulations , we adopted the mass of the host star to be @xmath72 @xmath73 for fit 1 , and @xmath74 @xmath75 for fit 2 . under the assumption of sin__i _ _ = 1 , the minimum masses of two companions are employed throughout the paper unless we state elsewhere .
in addition , we utilized the time step to be one percent of the orbital period of the inner planet ( hd 82943 c ) in the integrations .
the numerical errors were effectively controlled during the integration , with the local truncation error @xmath76 for the time span of 10@xmath77 yr , in the meantime the accuracy was also checked by the energy errors . in comparison , we also employed symplectic integrators ( feng 1986 ; wisdom & holman 1991 ) to integrate the same orbit to assure the results for some cases . in the research of the planetary systems , the maximum lyapunov characteristic exponent ( lce ) is usually adopted to identity the regular or chaotic orbits , since the chaotic orbits are sensitively dependent on the initial conditions , corresponding to a positive lce in the secular evolution . in this paper , we refer to the stability for the system means that the orbiting planets can remain periodic or quasi - periodic motions with bounded trajectories after the investigated time was done .
we define the unstable orbits that either of the planets is ejected far away or moves too close to their parent star , and in our simulations , the integration was automatically ended when meeting the following criteria : ( 1 ) either of the eccentricities approaches unity , ( 2 ) either of the semi - major axes exceeds the factor of 2 or reduces half of the starting values , ( 3 ) either of the planets collides with the star or these planets do each other when entering the scope of the mutual hill sphere .
to begin the study of the stable resonant geometry in hd 82943 , let us firstly recall several known facts in our solar system .
one may be aware that the pair of jupiter and saturn is in a near 5:2 commensurability .
in general , the mean motion resonance takes place in the pairs of the moons of the major planets ( such as the galilean satellites of jupiter : io - europa - ganymede ; see lee & peale 2002 ) , the asteroidal belt and kuiper belt objects ( kbos ) ( duncan , levison , & budd 1995 ; wan & huang 2001 ) . as more and more
extrasolar planetary systems are being discovered , the resonant pairs are frequently found to occur in the multiple systems , e.g. , hd 82943 ( gozdziewski & maciejewski 2001 ) , gj 876 ( lee & peale 2002 ) , and possibly hd 160691 ( gozdziewski , konacki , & maciejewski 2003 ; bois et al .
2003 ) in a 2:1 mmr , and 55 cnc in a 3:1 mmr ( ji et al .
2003a ) , and so on .
several theoretical or numerical works have been done to enhance the comprehension of the evolution into the resonances by non - conservative tidal forces ( ferraz - mello , beauge , & michtchenko 2003 ) , the convergent migration for the planets due to the disk - planet interaction by a damping force ( snellgrove et al .
2001 ; lee & peale 2002 ; kley 2003 ) , the excitation of the orbital eccentricities by repeated crossings of two companions migrating on divergent orbits ( chiang et al .
2002 ; chiang 2003 ) or the eccentricity growth by two resonant planets through inward migration ( murray , paskowitz , & holman 2002 ) , and resonant inclination excitation mechanism for migrating planets in the gas disk ( thommes & lissauer 2003 ) . on the other hand , if the two planets are trapped into a 2:1 resonance via one of the potential mechanisms , it is still necessary to understand their actual situation of orbital motions after the capture , and to investigate whether they can secularly survive in the system in some appropriate geometry and examine how the stability depends on the orbital parameters . in the usual notation of celestial mechanics , the lowest order eccentricity - type ( see murray & dermott 1999 ) resonant arguments @xmath78 , @xmath79 for the 2:1 mmr are @xmath80 @xmath81 where @xmath82 , @xmath83 are , respectively , the mean longitudes of the inner and outer planets , and @xmath84 , @xmath85 denote their apsidal longitudes , respectively . additionally , the relative apsidal longitudes of two companions @xmath86 reads , @xmath87 as we mentioned previously , the apsidal alignment ( or antialignment ) is found to exist in most of the multiple systems ( kinoshita & nakai 2000 ; rivera & lissauer 2000 ; chiang , tabachnik , & tremaine 2001 ; laughlin et al .
2002 ; lee & peale 2002 , 2003 ; gozdziewski et al . 2003 ; ji et al .
2003a , b ; zhou & sun 2003 ) , to play a significant role in retaining the stability of the system ( gozdziewski 2003 ; ji et al . 2003b ) . moreover
, this apsidal phase - locking of two orbits indicates that a pair of planets have common time - averaged rate of apsidal precession , and in the literature , it is also referred to apsidal resonance or corotation ( chiang et al . 2001 ; malhotra 2002 ) . for a 2:1 mmr
, it is easily noticed that for the above three arguments , no more than two are linearly independent , indicating that either all three librate or only one librate in the occurrence of the libration of the critical argument . in more general case ,
nelson & papaloizou ( 2002 ) also showed that both situations can occur for a @xmath88 eccentricity - type commensurability ( murray & dermott 1999 ) , where the associated resonant angles are defined by @xmath89(@xmath90 are positive integers and @xmath91 ) in the numerical surveys of the pair of the extrasolar planets .
as a matter of fact , this was again verified in the case of the 3:1 mmr for 55 cnc ( ji et al .
2003a ; zhou
et al . 2004 ) , where not only do all three resonant angles of @xmath92 , @xmath93 and @xmath94 ( here the subscript b , c , respectively , denote the two inner planets of 55 cnc b and 55 cnc c ) librate , but the relative apsidal longitude of @xmath95 librates about a constant angle simultaneously . in the simulations of hd 82943
, we found that three types of stable orbits can survive in this system : ( i ) only @xmath78 librates about @xmath69 , while @xmath79 and @xmath86 circulate ; ( ii ) case of alignment , three arguments all librate about @xmath69 , denoting by @xmath2 ; ( iii ) case of antialignment , both @xmath78 and @xmath86 librate about @xmath96 , while @xmath79 librates about @xmath69 , denoting by @xmath3 , @xmath4 , @xmath5 . in the following sections
, we will separately discuss above three stable configurations with both the numerical outcomes and analytical means and also compare them with remarkable evidences in other multiple systems involved in 2:1 resonance .
first , let us suppose two planets with the masses of @xmath97 and @xmath15 orbit the central star with the mass of @xmath98 .
here we only take into account the point - mass interaction between star - planet and planet - planet , without the consideration of the effects of the oblateness and general relativity arising from the parent star .
thus , in jacobi coordinates ( @xmath99 ; assume @xmath100 ) , the hamiltonian for three - body system can be written ( brouwer & clemence 1961 ) : @xmath101 where @xmath102 and @xmath103 here @xmath104 ( or @xmath105 ) is the distance between @xmath98 ( or @xmath97 ) and @xmath15 , @xmath106 the distance of @xmath15 with respect to the center of mass for @xmath98 and @xmath97 .
therefore , the disturbing potential @xmath107 can be further expanded to the following forms for the coplanar case , @xmath108 where @xmath109 here @xmath110 in ( [ eq7 ] ) , @xmath111 depends on the semi - major axes ( in association with the laplace coefficients ) and the eccentricities of two planets and this expansion of the disturbing potential also requires the smaller values of @xmath36 ( i=1,2 ) and the two orbits without intersection . in the case of gj 876 , lee & peale ( 2002 ) expanded ( [ eq7 ] ) and collected the terms up to @xmath112 , the truncation series may be valid for low values of @xmath36 , where @xmath113 and @xmath114 ( laughlin & chambers 2001 ) , without the occurrence of the crossing orbits . in a more recent work ,
beauge & michtchenko ( 2003 ) also developed an expansion method that can deal with higher values of the eccentricities up to a limit of @xmath115 , but they suggested that it is still related to the situation of the proximity between the planets .
however , as for hd 82943 , the planetary eccentricities can be as high as 0.41 - 0.54 for fit 1 , or 0.18 - 0.38 for fit 2 , the development of ( [ eq7 ] ) with respect to large eccentricities can not be exactly applied , because the expanding series , consisting of a great number of the expansion items , either converge very slowly or become divergent in the collision circumstances .
therefore , herein we seek for a semi - analytical treatment ( kinoshita & nakai 2002 ) for this question and then compare the analytical results with direct numerical integrations .
we emphasize that this semi - numerical method corresponds to a first - order theory with respect to the small parameter @xmath116 , where @xmath116 represents the mass ratio of the jupiter - like planet to its parent star and is normally of the order of @xmath117 in the planetary systems , and can apply to the planetary system with any eccentricities , even for the mutual crossing case .
the hamiltonian for the coplanar case is @xmath118 in order to keep the hamiltonian form , we should use jacobi coordinates to study this system . as the indirect part of the hamiltonian @xmath119
does not contribute to the secular part , we simply take the direct part . as the two orbits of hd 82943
are close to each other and may intersect ( ji et al . 2003c ; hadjidemetriou & psychoyos 2003 ; lee 2004 ) , the ordinary analytical expansions of the main part invalidate in the practical usage , then we adopt the original form of the main part of the disturbing function and numerically evaluate it .
again , by eliminating short - periodic terms in the classical average process , the new hamiltonian reads : @xmath120 the degree of freedom of the new hamiltonian ( [ eq9 ] ) is reduced from four to two . however , the hamiltonian @xmath121 is not integrable . as we shall see in 3.2 that the semi - major axes of the two resonant planets slightly change when they are close to 2:1 mmr ,
so we can assume that @xmath122 and @xmath123 are almost constant ( see figures 2 and 3 ) for the critical argument @xmath124 or @xmath125 .
in final , the degree of freedom of the new hamiltonian is reduced to one : @xmath126 we again eliminate @xmath127 in equation ( [ eq10 ] ) with conservation of the total angular momentum @xmath128 ( see also appendix ) , then we have @xmath129 thus we can draw the level curves of the hamiltonian and understand the global behavior of @xmath130 and @xmath131 . as the eccentricity @xmath130 of the inner planet becomes large , we numerically averaged the original hamiltonian ( [ eq8 ] ) under the condition of the critical argument @xmath124 or @xmath125 and the angular momentum conservation relationship , and then obtain numerically the averaged hamiltonian ( [ eq13 ] ) with the parameter of the total angular momentum @xmath128 .
the technique of the numerical averaging is extensively described in the paper ( kinoshita & nakai 1985 ) , in which the case of @xmath132 is also discussed . to make a study of the evolution of the eccentricities and the relative periapse arguments
, we can draw the contour map of hamiltonian ( [ eq13 ] ) by taking @xmath131 as the horizontal axis and @xmath130 or @xmath127 as the vertical axis ( see figure 4 ) with the parameter @xmath128 , which is determined from the initial conditions .
for the first 100 runs for fit 1 , each orbit integration lasts for @xmath133 yr .
in the simulations , we notice that all the stable cases are involved in the 2:1 mmr and easily understand that the stability of a system is sensitive to its initial planetary configuration .
furthermore , we noticed that 2% , 3% , and 2% of the cases belong to types i , ii , and iii and 93% of the systems destabilize , which means that either of the two planets leave _ in situ _ to the infinity with a rapid increase of the semi - major axes or the eccentricities grow much larger on the timescale @xmath134 yr or even shorter , owing to the mutual interaction between star - planet or planet - planet through frequent close approaches . in table 1 , are listed two sets of orbital parameters
fit 1a is related to the antialigned configuration for type iii and fit 1b is involved in the case for aligned orbits for type ii . here
the adopted parameters for two planets are the masses of 1.63 - 0.88 @xmath135 , the semi - major axes of 1.16 - 0.73 au , the eccentricities of 0.41 - 0.54 .
let us first examine the stable orbits of type i , where in the numerical simulations , we find that only one argument of @xmath78 librates about @xmath69 and the others @xmath79 and @xmath86 take up the full circulation from @xmath69 to @xmath70 . the semi - major axes @xmath136 and @xmath137 perform tiny oscillations about 0.73 and 1.15 au for the whole integration time
. type i was initially introduced by gozdziewski & maciejewski ( 2001 ) with the previous fits . however , in this paper , types ii and iii , which can grasp the couple of companions escaping from deviating their tracks in two directions of remaining both the semi - major axes and the eccentricities , are new findings of stable orbits for this resonant system .
hence , in the following sections , we primarily pay attention to the stable configurations linked to both a 2:1 resonance and the apsidal corotation .
figure 2 shows the typical orbital evolution for fit 1a ( antialigned orbits ) . in the figure
, we can see that the semi - major axes @xmath136 and @xmath137 slightly vibrate about 0.73 and 1.15 au for @xmath138 yr yr using symplectic integrator for this case , and no sign shows that the system will become chaotic for longer evolution . ] ( in the figure , to see more clearly , we simply display a snapshot for @xmath139 yr ) , meanwhile @xmath78 librates about @xmath96 with a moderate amplitude of @xmath140 , @xmath79 about @xmath69 with a smaller amplitude of @xmath141 and @xmath86 ( by _ thick line _ ) about @xmath96 with an amplitude of @xmath142 .
it is noteworthy to point out that the 2:1 resonance variable @xmath78 librates in accord with the relative apsidal longitudes @xmath86 , occupying the same period of @xmath33 300 yr .
this phenomenon can be easily understood because the above three angles are not independent in the course of the orbital motion .
thus we can call @xmath78 and @xmath86 as two fundamental variables to characterize the stable planetary geometry in 2:1 mmr .
in addition , the modulations of the semi - major axes @xmath143 and @xmath144 ( @xmath39 ) reveal the fact that the two planets are indeed trapped in a 2:1 mmr for the secular orbital evolution . in figure 2 , we further notice that the eccentricities of @xmath62 and @xmath63 separately range in ( 0.36 , 0.56 ) and ( 0.39 , 0.49 ) , bearing the equal libration periods of @xmath78 and @xmath86 .
the orbital antialignment of axes in hd 82943 is reminiscent of that in the 2 : 1 orbital resonances for io - europa system ( lee & peale 2002 ) , in which @xmath78 ( involved in the io s longitude of periapse ) librates about @xmath69 , and @xmath79 ( involved in the europa s longitude of periapse , @xmath86 librate about @xmath96 , respectively .
the antialigned circumstance for io - europa is equivalent to type iii in the sense of dynamically stable configuration , however , these jovian moons are occupying almost near - circular orbits resulting from the primordial or the tidal origin ( peale 2003 ) and the resonance configuration may come into reality when the eccentricities are small .
by contrast , the hd 82943 system differs the io - europa pair in that both of the planets possess high eccentricities at present day .
hence , the origin of the 2:1 resonance for hd 82943 ( fit 1a ) orbiting two massive planets with elliptic trajectories should be explained by an innovative mechanism rather than a capture induced by co - orbital differential migration ( s. peale 2003 , private communication ; see also 5 ) and perhaps stem from gravitational scattering in a crowded system that augment their eccentricities , i.e. , for a system with 10 planets ( adams & laughlin 2003 ) , the scattering processes can yield the full extent of possible eccentricity where @xmath145 . recently , bois et al . (
2003 ) found that the stable configuration of the hd 160691 system is possibly involved in a 2:1 mean motion resonance combined with an anti - aligned in two apsidal lines , with higher eccentricity ( @xmath146 ) of the outer planet .
obviously , their research supported the stable geometry of type iii of hd 82943 , which also implies that this resonant topology may be suitable especially for the planets bearing larger eccentricities .
moreover , this antialigned resonant configuration for fit 1a means that the conjunctions can take place when the inner planet moves near apoapse and the outer planet is close to the periapse , and _ vice versa _ , which indicates that the two planets are far from each other during their orbital evolution and protected from frequent close encounters .
therefore , we can safely conclude that the stability of the hd 82943 system , which is related to the above coplanar configuration for fit 1a , can be simultaneously sustained by two dynamical mechanisms : the 2:1 mmr and the apsidal antialignment .
figure 3 exhibits the orbital evolution for fit 1b ( aligned orbits ) .
consequently , let us note that the semi - major axes @xmath136 and @xmath137 do not change dramatically but undergo small oscillations about 0.73 au and 1.15 au for @xmath138 yr ( a snapshot for @xmath147 yr ) , further the eccentricities @xmath62 and @xmath63 librate in the extent of ( 0.5 , 0.8 ) and ( 0 , 0.45 ) , respectively , in the behavior of the converse cycles , owing to the conservation of the total angular momenta for the system . here
@xmath78 librates about @xmath69 with a small amplitude of @xmath148 , but @xmath79 and @xmath86 ( by _ thick line _ ) individually librate about @xmath69 with a large amplitude of @xmath149 . in this resonant aligned geometry
, we again observe that @xmath78 , @xmath79 and @xmath86 share the common librating period of @xmath150 yr , in the same time , coupled with the librating period of the eccentricities of @xmath62 and @xmath63 .
in addition , it is worthwhile to point out that at the time that @xmath78 , @xmath79 and @xmath86 all approach zero ( where @xmath78 and @xmath86 start from the negative values to zero ) , the eccentricity of @xmath62 reaches the maximum @xmath151 and @xmath63 is close to the minimum @xmath152 , thus at the aligned conjunction point , the separation between two planets @xmath153 ( where @xmath136 and @xmath137 are almost preserved on the condition of 2:1 resonance locking ) can reach the maximal value and the maximum separation can prevent them from strong mutual perturbation when they rendezvous at the synodic location .
similar analysis can be applied for the case of fit 1a , where both @xmath78 and @xmath86 run through @xmath96 ( see figure 2 ) .
again , let us make a brief comparison with the circumstance for the 2:1 resonant system gj 876 , which is also found to be in aligned configuration for the two orbits ( marcy et al .
lee & peale ( 2002 ) further found that the amplitude of the librations of three arguments are not far from @xmath69 by using laughlin - chambers solutions and they even indicated that all the librations are remained even for the amplitudes of @xmath78 amounting up to @xmath154 for some cases . as for fit 1b , let us bear in mind the fact that the planetary eccentricities are much larger than those for gj 876 , although they may have similar planetary orbital geometry .
sequently , the conjunctions always take place when two planets move close to their the periapse longitudes and the alignment can be maintained over long - term orbital evolution .
this , _ de facto _ , reveals that type ii coplanar configuration may be possible for real observed systems of gj 876 and hd 82943 .
here we adopted the semi - analytical model introduced in 3.1 to draw the hamiltonian contour and then compare these results with those given by direct integrations over secular orbital evolution .
in the averaged hamiltonian after removing the short - periodic terms , the difference arising from the choice between the jacobi coordinates and astrocentric coordinates is the order of the perturbation ( @xmath155 ) , which could be ignorable .
therefore , we adopt the astrocentric elements given in table 1 to be jacobi elements in the averaged model . as a comparison ,
the initial orbital elements for numerical integration are also converted into the jacobi coordinates . in figure 4
, we indicate that the contour diagram of hamiltonian ( [ eq13 ] ) are plotted against various levels by changing @xmath62 , @xmath63 and @xmath156 , _ left panel _ represents the results for fit 1a , where the contour levels are exhibited by _
thin line _ and the numerical solutions shown by _ thick line _
( the innermost curve about marked _ plus _ ) , while _ right panel _ for fit 1b .
the @xmath11 figures show a good agreement between the numerical outcomes and the semi - analytical solutions for two fits .
it is worthwhile to mention that the closest level of the hamiltonian contour to the numerical solution set the stable boundary for the apsidal resonant geometry , inside which the regular orbital motion for two orbits can be found with qualified eccentricities and the relative apsidal longitudes , thus we may conclude that this semi - analytical means is effective to help predict stable orbital solutions ( e.g. , beauge et al . 2003 ) . on the other hand ,
the diagrams of the hamiltonian contour further show that the stability zones for the eccentricities @xmath62 and @xmath63 are connected with the librating amplitude of @xmath95 : the smaller stability zones of the eccentricities , the narrower modulations of the relative apsidal longitudes , and _ vice visa_. in addition , we calculated the equilibrium points for each fit : ( @xmath96 , 0.45 ) and ( @xmath96 , 0.46 ) , respectively , for fit 1a ; ( @xmath69 , 0.68 ) and ( @xmath69 , 0.29 ) , respectively , for fit 1b .
as for fit 1b , we observe that there should exist stable orbits about the equilibrium center for @xmath157 ) , together with high eccentricity @xmath62 above 0.60 , which satisfy the necessary conditions for two planets evolving into a 2:1 resonance ( kley , peitz , & bryden 2004 ; see also 5 ) .
again , we recall that the period of the 2:1 resonant variable @xmath158 is equal to that of the relative apsidal longitudes @xmath131 ( see figs . 2 and 3 ) , which can be clearly reflected in fig .
moreover , in a view of the numerical results , we can see that the eccentricities of both planets are well restricted with the small - amplitude libration of @xmath131 about @xmath96 for fit 1a ( antialigned ) and large - amplitude libration of @xmath131 about @xmath69 for fit 1b ( aligned ) , which can also be consistent with the results in figs . 2 and 3 . as aforementioned , different selections for the initial orbital parameters will lead to stable or unstable configurations for the hd 82943 system . by analyzing the starting orbital data
, we discover that those stable planetary orbits involved in type ii ( or iii ) require the initial @xmath158 and @xmath131 should be constructed to satisfy the conditions of being less than tens of degrees from @xmath69 ( or @xmath96 ) , implying that the linear combinations of the mean anomalies , ascending nodes and periapse arguments can be close to the 2:1 resonant geometry . as for the second 100 runs near the vicinity of fit 2
, we found there exist two kinds of stable geometry related to a 2:1 mmr for this new fit ( mayor et al .
2004 ) , in which 16% of the stable systems can last for @xmath0 yr ( where 14% of the stable cases is in aligned orbits of type ii and 2% with simply @xmath78 libration for type i ) , and 84% are found to self - destruct at @xmath159 yr .
the percentage of the survival orbits indicates that there is a likelihood for the two planets of this system to be aligned . in table 2 , are listed the orbital parameters in association with the alignment of two orbits .
let us recall the adopted planetary masses of 1.84 - 1.85 @xmath135 , the semi - major axes of 1.18 - 0.75 au and the eccentricities of 0.18 - 0.38 . in the following subsections ,
the additional computations are carried out most frequently according to the initial orbits from fit 2 , and the typical secular evolution timescale ranges from @xmath160 yr to @xmath0 yr .
figure 5 illustrates the typical aligned orbital evolution for fit 2 . in comparison with figure 3
, we notice that the semi - major axes @xmath136 and @xmath137 are almost unchanged but modulate about 0.75 and 1.18 au with small amplitude for @xmath138 yr , at this time , the amplitude of the oscillations for @xmath62 and @xmath63 are not so large and the eccentricities are just wandering in the span ( 0.34 , 0.44 ) and ( 0 , 0.25 ) , respectively . here
@xmath78 librates about @xmath69 with a moderate amplitude of @xmath161 ( in coincidence with one of the aligned case for gj 876 found by lee & peale 2002 ; see also
3.2.2 ) , but @xmath79 and @xmath86 ( by _ thick line _ ) individually librate about @xmath69 with large amplitude of @xmath162 .
again , we notice that the eccentricities have the libration periods of @xmath33 600 yr , which are coupled with those of @xmath78 and @xmath86 .
it should mention that although the various starting orbital parameters between fit 2 and fit 1b can result in variational librating amplitudes in the eccentricities and resonant arguments , however , from a viewpoint of topology , there is not so much difference between two aligned orbital solutions .
furthermore , in figure 6 , we still plot @xmath11 hamiltonian contour ( by _ thin lines _ ) and the numerical results are presented by dotted points . the equilibrium solutions ( by _ plus
_ ) are : ( @xmath69 , 0.40 ) and ( @xmath69 , 0.13 ) , respectively .
the semi - analytical contour chart exhibits a good agreement with the numerical investigations of maintaining the eccentricities with larger @xmath86-libration amplitudes . on the other hand
, the figure implies that the @xmath86-libration will be broken up if the libration amplitudes exceed or approach the critical value of @xmath163 near the separatrix , then two successive scenarios may occur , either aligned orbits evolve into type i geometry that simply @xmath78 librates , or the system destabilize on the shorter timescale . in the sense of the present solutions , the aligned configuration for hd 82943 is not robust than the antialigned case , where a wider relative periapse longitudes can be initially adopted ( see figure 4 ) .
bois ( 2003 , private communication ) also confirmed that the stable strip of the alignment for the initial @xmath164 is much narrower than that of the antialignment in their numerical calculations for hd 82943 .
however , if the aligned orbits are close to librating center , the robustness can be strengthened with a different story . in order to distinguish the dynamical behaviors in orbital parameter spaces and explore how the stability zones dependence on these parameters
, we extensively perform abundant integrations for each series ( where in each series , we only allow one parameter vary but others keep unchanged ) to numerically search for the stable solutions . in the following sections , we will discuss these results respectively .
figure 7a show the numerical scanning of the hd 82943 system in the semi - major axes space of [ @xmath165 .
all other orbital parameters are taken from table 2 . in this group of the simulations , we choose @xmath166 [ 0.60 au , 0.80 au ] and @xmath167 [ 1.00 au , 1.30 au ] with the resolution of @xmath168 grids , each direction spaced 0.01 au . in fig .
7a , the cross indicates the unstable orbits and the filled circles for stable orbits that the planets survive at least @xmath160 yr . here
, we notice that there exists a stable strip with the width from @xmath169 au to @xmath170 au , indicating that these stable solutions are quite close to the 2:1 resonance .
meantime , the intersection between the vertical and horizontal lines represents for the present solution from fit 2 , which is well located in the stable map . in the case of 2:1 resonance
, we have @xmath171 , where @xmath172 , if both of orbits contract or expand @xmath173 times of the initial configuration ( e.g. , fit 2 ) and the planetary masses becomes @xmath174 of the initial masses ( where other orbital parameters remained ) , the semi - amplitude of the radial velocity will not change , that s the reason why we can see a linear stable ribbon in fig .
7a . in this sense , the semi - major axes can not be uniquely determined unless their orbital periods are previously obtained from the observations
. mayor et al .
( 2004 ) pointed out that one should be careful of employing the newly updated orbital fits ( fit 2 ) due to the derived best - fit solutions of lack of considering the planet - planet interaction .
hence , the exploration for the orbital parameter space is extremely useful to help understand the stability regions relying on the eccentricities .
similar to the numerical scanning in the semi - major axes space , here only both of the eccentricities of @xmath62 and @xmath63 are allowed to be free parameters . in this series of the investigations
, we let @xmath175 ( 0.0 , 0.60 ) and @xmath176 ( 0.0 , 0.60 ) and the resolution in [ @xmath62 , @xmath63 ] space is @xmath178 points , spaced 0.03 in each grid ( see figure 7b ) , the integration time is @xmath160 yr for each orbit . in the figure ,
the symbols are defined as in figure 7a , where the filled circles mean those orbits are steady over secular evolution . according to the eccentricity of the outer planet @xmath63 , we summarize the results as follows : ( 1)strongly stable cases , where @xmath16 ( 40% ) , ( 2)partially stable cases , where @xmath179 ( 16% stable , 9% unstable ) , ( 3)unstable cases , where @xmath180 ( 35% ) . as for case ( 1 )
, we should mention two points : firstly , the eccentricities of fit 2 ( shown by the crossing of two lines ) just reside in the stable map of [ @xmath62 , @xmath63 ] ; secondly , in this figure , we notice that there exist stable solutions when @xmath181 and @xmath182 , which is consistent with mayor et al .
( 2004 ) , who also provided the other solution involved in an aligned configuration with a near - circular orbit for outer planet but found the residuals do not dramatically change .
we can also note that a higher @xmath62 can be possible in this system .
both of cases ( 1 ) and ( 2 ) contain more that 50% of the stable orbits in the numerical scanning , exhibiting that only a moderate @xmath63 will favor the stability of the hd 82943 .
the case ( 3 ) shows that when @xmath63 becomes larger than 0.39 , no regular orbits will appear even though @xmath62 can be adopted to be small values , which suggests that in this situation the stability for hd 82943 is more sensitive to the orbital motion of the outer planet , which can greatly determine whether the system is in chaotic or regular status .
figures 7c and 7d exhibit our numerical exploration in the parameter spaces of [ @xmath183 and [ @xmath184 for the time span of @xmath160 yr , respectively , with the resolution of @xmath185 grids . the stable map in the [ @xmath183 space again
show the condition of the 2:1 mmr , where there is a linear relationship @xmath186 .
therefore , on the basis of fig . 7c , we can not only verify our original mean anomalies of fit 2 ( by intersection ) but find the other arguments related to stable orbital motions for two planets .
the stable diagram in the [ @xmath187 space reveals the fact that both of the apsidal longitudes of @xmath84 and @xmath85 ( where nodal longitudes are constant here ) precess at the equal rate about the star . the stable zones in fig .
7d are quite narrower , as we stated before , this results from the aligned configuration with a larger libration of @xmath188 ( ref .
fig . 5 ) .
both figures clearly illustrate that those stable configurations are linked to the 2:1 resonance and apsidal corotation . in this section ,
we expect to understand how the stability depends on the planetary mass ratio for the coplanar planetary configurations . here in our numerical study
, we still adopted the orbital elements from table 2 , except the masses of two planets @xmath97 and @xmath15 .
firstly , we varied @xmath189 in an incremental step of 0.05 ranging from 0.35 to 1.00 , and the rescaled masses of @xmath190 ( @xmath191 ) can conveniently obtained by the product of @xmath192 for a specific @xmath189 , where the mass ratio @xmath97/@xmath15 is fixed to be @xmath193 .
then 14 coplanar integrations were carried out for @xmath0 yr in this runs .
the numerical results show that the stable orbits ( of type ii ) for hd 82943 requires @xmath13 .
the stability set limitations on the planetary masses where @xmath194 $ ] and @xmath195 $ ] for different values of @xmath189 . in the case of gj 876 , similarly , marcy et al .
( 2001 ) reported that several stable systems , catalogued in type ii , had @xmath196 for the integration timescale up to 500 myr .
additionally , laughlin & chambers ( 2001 ) also suggested that @xmath189 possibly ranges from 0.5 to 0.8 for the coplanar fits for gj 876 . from above statements , we notice that @xmath13 is necessary for both systems occupying a 2:1 resonance and the extent of the planetary masses can further be estimated . in the other computations , we changed the mass ratio for @xmath97/@xmath15 and fixed the mass @xmath197 @xmath135 , let @xmath97/@xmath198 $ ] , then performed over thirty additional simulations to explore the stability of the system depending on the variational mass ratios . as a result , we found that @xmath97/@xmath199 is necessitated for the stable orbits , indicating an upper limitation for the mass of the inner planet when we remain that of the outer companion . in the following step , we replaced @xmath97 with much lower masses ranging from several @xmath200 to 20 @xmath200 for new investigations , and found that the stable geometry of type ii can still be remained and a surprise is that the eccentricity of the inner planet can amount up to @xmath201 with quasi - periodic oscillations .
these numerical explorations suggest that the apsidal configuration can maintain with much smaller masses for inner planet , which is consistent with the results given by beauge et al .
( 2003 ) , who analyzed the corotational solutions by gradually decreasing the mass of the inner planet and found that although the solutions simply depend weakly on the individual masses of the planets and the maximum value for the planetary mass of @xmath97 is indeed in existence with given eccentricities for the stable apsidal topology . in our solar system
, most of the major planets have small inclinations with regard to the essential reference plane .
however , there is a special case for the neptune - pluto system ( @xmath202 ) , where pluto orbit an eccentric trajectory with a large inclination of @xmath203 but neptune travels on a near - circular orbit with the inclination less than @xmath204 ( murray & dermott 1999 ) .
cohen & hubbard ( 1965 ) discovered that these two planets are involved in a 3:2 resonance with the argument @xmath205 ( @xmath206 denote , respectively , the mean longitudes of pluto and neptune , @xmath207 is the longitude of perihelion of pluto ) .
these studies reveal that there may exist a relative inclination configuration between two orbits in association with the mean motion resonance . in this sense , we anticipate to understand the situation for the present exosystem .
we again performed extra tests of mutually inclined orbits for two resonant planets for the integration of @xmath71 yr , to examine the stability of the hd 82943 system .
the orbital data were taken from table 2 and for simplicity , we just adopt their minimum masses and remain them unchanged for our additional integrations . for the initials , here we simply altered the inclination of the outer planet @xmath208 and increased @xmath208 in an increment of @xmath209 from @xmath210 to @xmath211 , but always assumed the inclination of the inner planet @xmath212 to be @xmath61 .
the numerical integrations reveal the fact that although the inclination was greatly changed , the two planets are still in 2:1 mmr and apsidal alignment ( type ii ) with the inclination of @xmath213 . on the contrary
, for @xmath214 , we found that the eccentricity of the inner planet can pump up to unity due to the great perturbation exerted by the outer planet and leave the orbits eventually .
the relative inclination @xmath215 can be determined by @xmath216 , where @xmath217 and @xmath218 are , respectively , the longitudes of ascending nodes for two planets . to sum up , in the case of the non - coplanar configuration for hd 82943
, stability requires that the relative inclination be @xmath12 or less . in the case of gj 876 ,
rivera & lissauer ( 2001 ) carried out dynamical fits for the possible best solutions for the mutual inclination geometry and found that the stability for gj 876 requires the relative inclination to be @xmath219 or less .
therefore , it should be mentioned that the mutually inclined orbits dynamically survive over long - term timescale in the planetary geometry locking into a 2:1 mmr , but it should place constraints on the initial conditions for two planets .
thommes & lissauer ( 2003 ) showed that two planets migrating in resonance can perform significant excitation of their inclinations and the relative inclination between two planets can grow to @xmath140 in their three - dimensional simulations . as a final note for this part ,
the above - mentioned numerical investigations for various mutual inclined exosystems show that the low relative inclination do play an important role in the stability for such systems , and this mutual inclined configuration may arise from the inclination - type resonance ( thommes & lissauer 2003 ) associated with the longitudes of the ascending nodes for inner and outer planets if the inner planet owns a high eccentricity ( e.g. , for fit 2 , @xmath220 ) , while the higher relative inclination geometry can destabilize the system ( e.g. , lissauer & rivera 2001 ) on shorter timescales .
as we mentioned previously , several studies have been devoted to explore the orbital motions of the exosystems in 2:1 mmr . in the case
gj 876 , lee & peale ( 2002 ) discovered that there are stable configurations with @xmath78 , @xmath79 and @xmath86 all librating about @xmath69 ( related to type ii ) for @xmath221 , and kinoshita & nakai ( 2001 ) and ji et al .
( 2002 ) also studied this system and confirmed the 2:1 resonance acting as an effective mechanism to hold the system , moreover , an extended stable geometry of type i was shown by ji et al .
( 2003b ) for gj 876 .
in other works , hadjidemetriou & psychoyos ( 2003 ) numerically studied the families of periodic orbits for hd 82943 in the rotating frame , and indicated that two kinds of families orbits can survive for the 2:1 resonant planets in their computations .
however , it is the first time that we show an exhaustive exploration in the orbital parameter space for the hd 82943 system based on the updated solutions , which reveals the new dynamical features for this system . in summary ,
the results presented in this article on the possible stable planetary geometry involved in a 2:1 resonance are not only consistent with the previous studies , but also outline an extensive dynamics of these systems .
moreover , we point out that the aforementioned three groups of resonant configurations imply a general principle that governs two planets in a dynamically stable state over secular orbital evolution in the planetary systems , which should be examined by more future discoveries of the 2:1 resonant pairs .
still , we will concentrate on whether a postulated terrestrial can survive in such geometry ( see 4 ) , and how a certain system is evolved into an apsidal alignment or antialignment resonant configuration in 5 .
the habitable zones ( hz ) are generally convinced to be suitable places for terrestrial planets that can provide the liquid - water , subtle temperature and atmosphere environment , and other proper conditions ( kasting et al .
1993 ) , supporting the development and biological evolution of life on their surfaces .
the habitable region is considered to be centered at @xmath222 = 1 au @xmath223 ( gould , ford , & fischer 2003 ) or @xmath224 1.03 au @xmath223 ( kuchner 2003 ) , such that a water planet can exist in equilibrium with stellar radiation , where @xmath222 is the radius of the planet s orbit and @xmath225 is the luminosity of the host star . in a viewpoint of the dynamics , first of all , there should be stable orbits at @xmath18 au for earth - like planets moving about main sequence stars over very long timescale and we can also refer this to the dynamical habitability . in a recent work ,
menou & tabachnik ( 2003 ) exhaustively studied the dynamical habitability of 85 known extrasolar planetary systems through numerical simulations of their orbital dynamics in the presence of potentially habitable terrestrial planets .
they found that more than half of the known exosystems can not harbor habitable terrestrial planets and about 25% of the systems mostly with close - in giant planets are dynamically habitable , similar to our solar system . and
in this section , we only draw our attention to the potential habitable zones for the systems ( e.g. hd 82943 and gj 876 ) with a stable geometry in a 2:1 resonance . here
we performed numerical surveys in examining the habitable zones both for hd 82943 and gj 876 . and at first , we considered stable configurations for hd 82943 ( fit 2 ) and gj 876 ( keck and lick fit with @xmath226 ; see table 2 in laughlin & chambers 2001 ) , which can be viewed as genuine systems closest to the actual observations .
then , we generated 100 seed planets that all bear the same masses of earth ( where @xmath227 ) in the simulations for each system .
in figure 8 , we can see the distribution of the initial orbital elements for these postulated planets
they initially move on the less inclined belt with the relative inclination with regard to the plane of the resonant pair no more than @xmath209 , the semi - major axis @xmath228 [ 0.96 au , 1.05 au ] , the eccentricity @xmath229 $ ] , and the leftover arguments are set at random .
here we simply accounted for the near - circular orbits rather than much more eccentric orbits , because the life should develop and evolve in the biological environment without larger variations of temperature , implying the earth - like planet should not move neither too close nor too far from the parent star . for each star - two - planet-``earth ''
four - body system , we carried out the integration for the timescale of @xmath0 yr , to examine the dynamical habitability for the hd 82943 system . as we shall see
, we find that none of the orbits can retained stable in the region at @xmath18 au for the integration time , the seed planet travels on hyperbolic or parabolic orbits at the characteristic timescale @xmath230 yr ( where 81% of them are ejected at @xmath231 yr and 95% at @xmath232 yr , see figure 9a ) , owing to the scattering ( lin & ida 1997 ) between two close resonant planets for the excitation in the eccentricity .
a typical orbital evolution for the ejected orbits is shown in figure 9b , the semi - major axe @xmath30 grows from @xmath18 au to over 100 au , the eccentricity @xmath31 undergoes a rapid increase from @xmath233 to 1 and the inclination @xmath32 is also excited to a high value of @xmath234 ( in some cases , the inclination can exceed @xmath163 ) , finally , at @xmath235 yr , the assumed earth - like planet is scattered away from the two - planet system , while the orbital motions of the resonant planets remain as usual .
the gravitational scattering on the terrestrial planet caused an unstable motion either as an ejection or a catastrophic collision with the star or planets due to the dynamical instabilities ( ford , havlickova , & rasio 2001 ) . our numerical investigations suggest that there will be least possibility for stable earth - like orbits surviving in the range of [ 0.73 au , 1.18 au ] , even for wider habitable regions [ 0.75 au , 1.40 au ] given by kasting et al .
( 1993 ) , which is in good agreement with the simulations by menou & tabachnik ( 2003 ) .
analytically , gladman ( 1993 ) attained the minimum separation @xmath236 between two planets moving on initially circular orbits require for stability , @xmath237 where the mutual hill radius @xmath238 is @xmath239 for fit 2 , we achieve @xmath240 au and @xmath241 au , which these values may become a bit different as the two planets for hd 82943 orbit on eccentric paths .
nevertheless , our goal here is to present a qualitative analysis , and we can easily notice that the initial orbits for the assumed planets are quite close to hill sphere regime , resulting in their unstable motions .
but it can not rule out other stable regions for additional companions .
for example , sandquist et al . (
2002 ) explored a three - planet case for hd 82943 , which a supposed jovian planet with the mass of 0.5 @xmath135 at 0.02 au was added to the system , and they found that this additional planet does not affect the orbital motion of the two outer planets and also not disturb the stability of this system .
on the other hand , for the exterior domain extending to @xmath242 au @xmath243 au or even farther out , does there exist the dust rings or circumstellar disk that are analogous to the kuiper belt being a remnant of the solar nebula ?
the recently launched sirtf mission will provide fruitful information on this point in its ongoing tasks .
as a comparison , we accomplished the other same runs for gj 876 .
and we conducted the four - body integrations for the time span of integration of @xmath160 yr ( @xmath244 orbital periods for the innermost planet ) and found all the systems are stable for the integration timescale .
figure 10 exhibits the typical orbital evolution for the earth - like planet , both the semi - major axis @xmath30 and the eccentricity @xmath31 execute small fluctuations about 1 au and 0.06 , respectively , and the inclination @xmath32 also remains less than 2 degrees in the same time span .
and there is no sign to indicate that such regular orbits at @xmath18 au with low eccentricities will become chaotic for much longer time even for the age of the star . due to the faint luminosity for m4 star gj 876
, the habitable zones may reside in the range of [ 0.10 au , 0.20 au ] ( kasting et al .
1993 ; menou & tabachnik 2003 ) sufficiently nearer to the orbits of gj 876 b ( @xmath33 0.13 au ) and gj 876 c ( @xmath33 0.21 au ) ( marcy et al . 2001 ; laughlin & chambers 2001 ; rivera & lissauer 2001 ) , hence , the global chaos , corresponding to the circumstance of a collisional disruption or ejection , can always take place according to the hill stability criterion ( gladman 1993 ; chambers , wetherill & boss 1996 ) or direct numerical examinations , and exclude the possible existence of earth - like planets for such habitable regions . whereas kuchner ( 2003 ) pointed out that a liquid planet with @xmath224 1.03 au @xmath223 ( where for gj 876 , @xmath245 au ) may exist under the protection of a thick steam atmosphere , it seems plausible for the earth - like planets revolve around gj 876 at the orbit about 1 au , in the concept of the dynamical habitability .
however , our predictions should be confirmed through careful investigation in future research if the earth - like planets do survive secularly in such habitable zone .
firstly , the direct evidence of the presence for such planets strongly depend upon the future high resolution astrometric measurements , e.g. , space interferometer mission ( sim ) or terrestrial planet finder ( tpf ) that can provide much better precision than the present long - term precision of @xmath33 2 - 3 m s@xmath246 ( butler et al .
2003 ) for ground - based doppler surveys . secondly ,
the simulations of the runaway accretion of the planetary embryos ( laughlin et al .
2002 ) suggest that it is impossible for the earth - mass planets to be formed in habitable zones even for a well - separated system with near - circular orbits ( e.g. , 47 uma ) , because the formation of the terrestrial planets are terrifically constrained by giant planet migration ( armitage 2003 ) , therefore the depletion of small planetesimals in a disk - clearing phase may possibly take place when the massive planets sweep over them in their orbital decay process unless this orbital migration can halt at a proper position allowing for the continuous growth of the planetary embryos or the already - created earth - like planets can also follow the migration with the giant planets in step and cease about the habitable regions .
nevertheless , fortunately , the terrestrial planets are conceivable to survive in a modest fraction of systems where a single generation of massive planets formed ( armitage 2003 ) without significant migration .
the short - period giant planets that are close to the parent star with high temperature ( e.g. , 51 peg , so - called hot jupiter with @xmath247 au ) , are now thought to be formed through the shrinking orbital migration rather than be formed _ in situ _ due to the tidal interaction of a protoplanet with a surrounding gaseous disk ( ward 1997 ) and this orbital migration can originate from the exchange of angular momentum between the natal disk and protoplanet ( papaloizou 2003 and references therein ) .
now , it is widely believed that the migration will ubiquitously come about when two - planet systems suffer tidal interaction with an exterior disk , the outer planet will migrate inward until it gets trapped in a low order mean motion resonance with the inner companion ( bryden et al .
2000 ; kley 2000 ) and the eccentricities will stop growing and be balanced as steady equilibrium values ( nelson & papaloizou 2002 ) after being captured into the resonances . for the aligned orbits of type ii , in the earlier works for gj 876 , snellgrove et al .
( 2001 ) performed full hydrodynamical two - dimensional simulations to study the inward migration for gj 876 planets interacting with each other and a protoplanetary disk , and the outer planet can be trapped into a 2:1 commensurability with the inner planet through slow migration , where the orbits were observed to be aligned with the resonant angles librating about zero ( see also an overview in papaloizou 2003 ) and the system can survive at least @xmath248 orbits by the removal of the external disk .
subsequently , lee & peale ( 2002 ) showed that the forced inward migration of the outer companion of the gj 876 system leads to certain capture into the observed resonances if the initial eccentricities are smaller enough where @xmath249 and @xmath250 with proper migration rates .
they also revealed that after resonance capture , the eccentricities increase rapidly due to resonant interactions between the planets and the forced migration without eccentricity damping , while with parameterized eccentricity damping where @xmath251 , the eccentricities can reach equilibrium values that remain unchanged for sufficiently long migration in the resonances .
recently , in the study of the evolution in resonant planetary systems ( e.g. , hd 82943 ) , kley et al .
( 2004 ) found that the two planets can enter into a 2:1 resonance in varying timescale provided that @xmath252 , with both full hydrodynamical simulations and damped n - body calculations .
they further suggested that there are several requirements for the 2:1 resonance capture the more massive mass for the outer companion , the higher eccentricity of the inner planet and the apsidal alignment of @xmath253 . in the case of type ii (
fit 2 ) , where the eccentricity @xmath254 and a moderate @xmath220 and with two nearly equal planetary masses , or in the case of fit 1b , where there can also be stable geometry about @xmath255 and a higher @xmath62 with a larger outer planet , in both cases the two planets undergo @xmath86-libration about @xmath69 , thus we may comment that if the hd 82943 system is most likely to be aligned ( lee m.h .
2003 , private communication ) , then the origin for such orbital geometry can be possibly attained by the predictions of the hydrodynamic models ( kley et al .
2004 ) , we will address this problem in our forthcoming paper .
for the antialigned orbits of type iii , thommes & lissauer ( 2003 ) argued that if the planets reached this configuration by migration in resonance , a significant mutual inclination between their orbits would seem to be required .
as in this geometry , the two orbits for hd 82943 can experience the intersection on the co - orbital region ( ji et al . 2003c ; lee 2004 ) , therefore a moderate inclination between two orbits can be achieved to act as a protection mechanism against close approaches ( thommes & lissauer 2003 ) . in 3.3.2 , we also show that the highly inclined resonant configuration for this system can also be in presence , which the system is stabilized by both the 2:1 resonance and the apsidal phase - locking .
however , for fit 1a , although the hd 82943 system can be stable at least for @xmath0 yr , it is still difficult to secure the resonant capture with n - body models due to the larger eccentricities of 0.41 - 0.54 in kley et al .
( 2004 ) , the origin for such configuration may attribute to an additional passing star s influence or pure gravitational scattering mechanism ( lin & ida 1997 ; adams & laughlin 2003 ) , to pump up the eccentricities .
one of the final fates in the scattering procedure is that the inner planet is left alone in the system with an eccentric orbit while the outer planet is ejected ( ford et al .
2001 ) into the stellar space , nearly immune to the gravitational influence from the parent star .
if the scattering does happen in the course of the migration , a casual rapid capture may be required to prevent one of the planets from throwing away from the system and subsequently two orbits are well locked into 2:1 resonance ( see 3.2.1 ) at larger eccentricities .
alternatively , another possible scenario is that both of the eccentricities are excited to moderate values when the 2:1 resonance is being crossed ( chiang et al .
2002 ) , subsequently the continuous resonance crossings may terminate for the suitable eccentricities evolved into the resonance geometry .
in this paper , we have explored the stable geometry for a system with two planets involved in a 2:1 mmr , and we mainly concentrate on the study of the hd 82943 system by adopting two different sets of the orbital parameters . in our numerical simulations , we found there are three possible stable configurations for hd 82943 system:(1)type i , only @xmath1 , ( 2)type ii , @xmath2 ( aligned case ) , and ( 3)type iii , @xmath3 , @xmath4 , @xmath5 ( antialigned case ) .
the direct integrations show that all stable orbits are related to the 2:1 commensurability that remains the semi - major axes for two companions slightly vary about the initial values over the secular dynamical evolution for @xmath0 yr , further the apsidal phase - locking for two planets can enhance the stability for this system , because the eccentricities are simultaneously preserved to prevent the planets from frequent close encounters . additionally , we comment that other systems trapped in 2:1 mmr ( e.g. gj 876 or possibly hd 160691 ) can select one of the aforementioned three stable topology in their authentic orbital motions , which are determined by their initial parameters . and we underline that such configurations , making up the dynamical families , can serve as a general regulation for the long - term stable orbits of the 2:1 resonant exosystems . in final , we summarize some conclusions : using the earlier best - solutions of fit 1 , we found three types of stable orbits for hd 82943 , while we observed the steady configurations of types i and ii by employing the new solution of fit 2 supplied by mayor et al .
( 2004 ) . in the meantime
, we also proposed a semi - analytical model to study the @xmath11 hamiltonian contour determined by the initial parameters , and the closest level curves encompass the direct numerical results , which presents a good agreement between them .
the theoretical figures can still provide valuable information on the dynamics for two planets . given the stable geometry based on fit 2
, we then extensively examined the dependence of the stability of hd 82943 in the orbital parameter space and the planetary mass ratios . in the case of the non - coplanar circumstances
, we found that stability requires that the relative inclination be @xmath12 or less .
for a fixed planetary mass ratio @xmath18 , the stable orbits for hd 82943 requires @xmath13 . for the non - constant mass ratio case ( where @xmath15 is kept ) ,
the requirement of the stability is @xmath97/@xmath199 , which indicates an upper boundary for the mass of the inner planet .
concerning the eccentricities , the system can be always steady when @xmath256 and @xmath17 . in a word , these outcomes do demonstrate that the stability for hd 82943 is strongly sensitive to the sound planetary masses , rather lower relative inclination between two orbits , the eccentricities and other orbital parameters .
in addition , we showed that the assumed terrestrial bodies can not exist in the habitable zones of hd 82943 thanks to strong perturbations induced by two massive planets , but the earth - like planets can be dynamically habitable in the gj 876 system at @xmath18 au in the numerical surveys .
however , as the additional measurements will improve the present best - fit solution for hd 82943 with a corrected model by taking into account the planet - planet interactions .
thus , we hope to have more precise orbital solution not only to check for our presented predictions in this paper but to better make sense of the complete dynamics and the origin of this system in the near future .
adams , f. , & laughlin , g. 2003 , icarus , 163 , 290 armitage , p. j. 2003 .
, 582 , l47 barnes , r . , & quinn , t. 2004 , , submitted , ( astro - ph/0401171 ) beauge , c. , ferraz - mello , s. , & michtchenko , t.a .
2003 , , 593 , 1124 beauge , c. , & michtchenko , t.a .
2003 , , 341 , 760 bois , e. , et al .
2003 , , 598 , 1312 brouwer , d. , & clemence , g. m. 1961 , methods of celestial mechanics ( new york : academic ) bryden , g. , et al .
2000 , , 540 , 1091 butler , r. p. , et al .
2003 , , 582 , 455 callegari jr . ,
n. , michtchenko , t.a . , & ferraz - mello , s. 2004 , cel .
mech . & dyn .
astron . , 89 , in press chambers , j. e. , wetherill , g. w. , & boss , a. 1996 , icarus , 119 , 261 chiang , e.i.,tabachnik , s . , & tremaine , s .
2001 , , 122,1607 chiang , e.i . , fischer , d. , & thommes , e .
2002 , , 564 , l105 chiang , e.i .
2003 , , 584,465 cohen , c.j . , &
hubbard , e.c .
1965 , , 70 , 10 duncan , m.j . ,
levison , h.f .
, & budd , s.m .
1995 , , 110 , 3073 dvorak , r. , et al .
2003 , , 410 , l13 feng k. 1986 , j. comp .
math.,4 , 279 ferraz - mello , s. , beauge , c. , & michtchenko , t. 2003 , cel .
mech . & dyn .
astron . , 87 , 99 fischer , d. , et al .
2001 , , 551 , 1107 fischer , d. , et al .
2003 , , 586 , 1394 fischer , d. a. , valenti , j. , & marcy , g. 2004 , in asp conf .
s219 , stars as suns : activity , evolution and planets , ed .
a. dupree and a. benz , ( san francisco : asp ) , in press ford , e. b. , rasio , f. a. , & sills , a. 1999 , , 514 , 411 ford , e. b. , havlickova , m. , & rasio , f. a. 2001 , icarus , 150 , 303 gladman , b. 1993 , icarus , 106 , 247 goldreich , p. , & tremaine , s. 1980 , , 241,425 gozdziewski , k . , &
maciejewski , a.j .
2001 , , 563 , l81 gozdziewski , k. 2003 , , 398 , 1151 gozdziewski , k. , & maciejewski , a.j .
2003 , , 586 , l153 gozdziewski , k. , konacki , m. , & maciejewski , a.j .
2003 , , 594 , 1019 gould , a. , ford , e. b. , & fischer , d.a .
2003 , , 591 , l155 hadjidemetriou , d.h .
. mech . & dyn .
astron.,83,141 hadjidemetriou , d.h . , &
psychoyos d. 2003 , lecture notes in physics , accepted haghighipour , n. , et al .
2003 , , 596 , 1332 israelian , g. , et al .
2001 , , 411 , 163 israelian , g. , et al .
2003 , , 405 , 753 jiang , i.g . , & yeh , l.c .
2004 , , submitted ji , j. h. , li , g.y . , & liu , l. 2002 , , 572,1041 ji , j. h. , et al . 2003a , , 585 , l139 ji , j. h. , et al .
2003b , , 591 , l57 ji , j. h. , et al .
2003c , cel .
astron . , 87 , 113 jones , b. , & sleep , p .
2002 , , 393,1015 jones , h.r.a . , et al .
2002 , , 337 , 1170 kasting , j. f. , whitmore , d. p. , & reynolds , r. t. 1993 , icarus , 101 , 108 kinoshita , h. & nakai , h .
2001,,53,l25 kinoshita , h. , & nakai , h .
1985 , cel .
mech . , 36 , 391 kinoshita , h. , & nakai , h .
2000 , in asp conf .
s202 , planetary systems in the universe , ed .
a. j. penny , p. artymowicz , a. -m .
lagrange , and s. s. russell ( san francisco : asp ) , in press kinoshita , h. , & nakai , h
. 2002,in proc .
34rd symp .
mech.,in press kiseleva - eggleton , l. , et al .
2002 , , 578 , l145 kley , w. 2000 , , 313 , l47 kley , w. 2003 , cel .
mech . & dyn .
astron . , 87 , 85 kley , w. , peitz , j. , & bryden , g. 2004 , , 414 , 735 kuchner , m.j .
2003 , , 596 , l105 laughlin , g. , & chambers , j . 2001,,551,l109 laughlin , g. , chambers , j. , & fischer , d. 2002 , , 579 , 455 lee , m.h . , &
peale , s.j .
2002 , , 567 , 596 lee , m.h . , & peale , s.j .
2003 , , 592 , 1201 lee , m.h .
2004 , , submitted , ( astro - ph/0401410 ) lin , d. n. c. , bodenheimer , p. , & richardson , d. c. 1996 , , 380 , 606 lin , d. n. c. , & ida , s. 1997 , apj , 477 , 781 lissauer , j. j. , & rivera , e. j. 2001 , , 554 , 1141 menou , k. , & tabachnik , s. 2003 , apj , 583 , 473 malhotra , r. 2002,,575,l33 mazeh , t .
, & zucker , s .
2003 , , 590,l115 marcy , g.w . ,
cochran , w. d. & mayor , m. 2000 , in protostars and planets iv , ed . v. mannings , a.p . boss & s. s. russell ( tucson : university of arizona press ) , p.1285 marcy , g. w. , et al .
2001 , , 556 , 296 marcy , g.w . , et al .
2003 , in scientific frontiers in research on extrasolar planets , ed .
d. deming and s. seager(washington dc : asp ) , vol.294 , p.1 mayor , m. , & queloz d. 1995 , , 378 , 355 mayor , m. , et al . 2004 , , 415 , 391 murray , c. d. , & dermott , s. f. 1999 , solar system dynamics ( new york : cambridge univ . press ) .
murray , n. , paskowitz , m. , & holman , m. 2002 , , 565,608 naef , d. , et al .
2001 , , 375 , l27 nelson , r. p. , papaloizou , j.c.b . ,
masset , f . , & kley , w. 2000 , , 318 , 18 nelson , r.p . , & papaloizou , j.c.b .
2002 , mnras , 333 , l26 papaloizou , j.c.b .
2003 , cel .
mech . & dyn .
, 87 , 53 peale , s.j .
2003 , cel .
mech . & dyn .
astron . , 87 , 129 rivera , e. j. , & lissauer , j. j. 2000 , , 530 , 454 rivera , e. j. , & lissauer , j. j. 2001 , , 558 , 392 santos , n. , israelian , g. , & mayor , m. 2001 , , 373 , 1019 santos , n. , et al . 2003 , , 398 , 363 sandquist , e.l . , et al . 2002 , , 572,1012 snellgrove , m.d . ,
papaloizou , j.c.b . , & r. p. , nelson .
2001 , , 374 , 1092 thommes , e. , & lissauer , j. j. 2003 , , 597 , 566 wan , x. s. , & huang , t. y. 2001 , , 368 , 700 ward , w. r. 1997 , icarus , 126 , 261 weidenschilling , s. j. , & marzari , f. 1996 , , 384 , 619 wisdom , j. , & holman , m. 1991 , , 102,1528 zhou , l. y. , et al .
2004 , , submitted,(astro - ph/0310121 ) zhou , j. l. , & sun y. s. 2003 , , 598 , 1290
the hamiltonian with use of jacobi coordinates has the following form : @xmath257 where @xmath258 now we have two choices for the unperturbed hamiltonian : + a ) @xmath259
b ) @xmath260 the difference between a ) and b ) is the last term . the choice a ) is recently popular . in the treatment of the triple star system
the choice b ) is always adopted , since the form of the disturbing function becomes simpler for the case @xmath261 .
the form of the total angular momentum @xmath128 is + a ) @xmath264 b ) @xmath265 in the above expressions the meaning of the osculating elements @xmath266 is different between a ) and b ) , because the unperturbed form is different .
however , for the averaged hamiltonian after the elimination of the short periodic terms , the difference due to the choice between the jacobi coordinates and heliocentric ( or astrocentric ) coordinates is the order of the perturbation ( @xmath267 ) , which could be negligible .
@xmath268 sin@xmath32(@xmath41 ) & 1.63 & 0.88 & & 1.63 & 0.88 + @xmath269(days ) & 444.6 & 221.6 & & 444.6 & 221.6 + @xmath30(au ) & 1.16 & 0.73 & & 1.16 & 0.73 + @xmath31 & 0.41 & 0.54 & & 0.41 & 0.54 + @xmath66(deg ) & 14.79 & 154.23 & & 327.65 & 239.11 + @xmath65(deg ) & 96.0 & 138.0 & & 96.0 & 138.0 + mean anomaly(deg ) & 97.86 & 10.02 & & 53.40 & 188.50 lcc @xmath268 sin@xmath32(@xmath41 ) & 1.84 & 1.85 + @xmath269(days ) & 435.1 & 219.4 + @xmath30(au ) & 1.18 & 0.75 + @xmath31 & 0.18 & 0.38 + @xmath66(deg ) & 120.91 & 315.60 + @xmath65(deg ) & 237.0 & 124.0 + mean anomaly(deg ) & 168.56 & 185.12 | we have numerically explored the stable planetary geometry for the multiple systems involved in a 2:1 mean motion resonance , and herein we mainly concentrate on the study of the hd 82943 system by employing two sets of the orbital parameters ( mayor et al .
2004 ) . in the simulations , we find that all stable orbits are related to the 2:1 commensurability that can help to remain the semi - major axes for two companions almost unaltered over the secular evolution for @xmath0 yr , and the apsidal phase - locking between two orbits can further enhance the stability for this system , because the eccentricities are simultaneously preserved to restrain the planets from frequent close encounters . for hd 82943
, there exist three possible stable configurations:(1 ) type i , only @xmath1 , ( 2 ) type ii , @xmath2 ( aligned case ) , and ( 3 ) type iii , @xmath3 , @xmath4 , @xmath5 ( antialigned case ) , here the lowest eccentricity - type mean motion resonant arguments are @xmath6 and @xmath7 , the relative apsidal longitudes @xmath8 ( where @xmath9 are , respectively , the mean longitudes of the inner and outer planets;@xmath10 are the longitudes of periapse ) . and we find that the other 2:1 resonant systems ( e.g. , gj 876 or hd 160691 ) may possess one of three stable orbits in their realistic motions .
in addition , we also propose a semi - analytical model to study @xmath11 hamiltonian contours , which are fairly consistent with direct numerical integrations . with the updated fit
, we then examine the dependence of the stability of this system on the relative inclination , the planetary mass ratios , the eccentricities and other orbital parameters : in the non - coplanar cases , we find that stability requires the relative inclination being @xmath12 or less ; as to the planetary mass ratio , the stable orbits for hd 82943 requires @xmath13 for a fixed value or @xmath14 where @xmath15 remains for the varying mass ratio ; concerning the eccentricities , the system can be always steady when @xmath16 and @xmath17 .
moreover , we numerically show that the assumed terrestrial bodies can not survive near the habitable zones for hd 82943 due to the strong perturbations induced by two resonant companions , but these low - mass planets can be dynamically habitable in the gj 876 system at @xmath18 au in the numerical surveys .
finally , we present a brief discussion on the origin of the 2:1 resonance for hd 82943 . |
the thermodynamics of black holes has been studied extensively since the work of hawking @xcite .
the notion of critical behaviour has arisen in several contexts for black holes , ranging from the hawking - page @xcite phase transition in hot anti - de - sitter space and the pioneering work by davies @xcite on the thermodynamics of kerr - newman black holes , to the idea that the extremal limit of various black hole families might themselves be regarded as genuine critical points @xcite . as for standard statistical mechanical systems ,
critical points are signalled by singularities in the specific heat .
more recently , various groups have investigated the application of ideas from information geometry to the study of black hole thermodynamics .
the use of information geometry @xcite in statistical mechanics in general was largely pioneered by ruppeiner @xcite and weinhold @xcite , who suggested that the curvature of a metric defined on the space of parameters of a statistical mechanical theory could provide information about the phase structure .
specifically , from consideration of fluctuations , ruppeiner suggested a metric based on the entropy @xmath0 where @xmath1 is the entropy , @xmath2 is the mass and @xmath3 are the other extensive thermodynamic variables of the theory under consideration .
it was found that the curvature of this metric was zero for a non - interacting theory such as an ideal gas , but non - zero for an interacting theory such as a van der waals gas , and divergent at the phase transition points @xcite .
the ruppeiner metric is conformally related to the weinhold metric @xcite by where @xmath4 is the temperature of the system under consideration .
this weinhold metric is defined as the hessian of the energy ( mass ) with respect to entropy and other extensive parameters , namely for non - black - hole systems , the results from using either metric have proved to be consistent @xcite , but consideration of different black hole families under various assumptions has led to numerous puzzling results for both metrics and inconsistencies between them @xcite .
wei _ et al .
_ @xcite have recently added to this catalogue of inconsistencies by examining the ruppeiner metric and curvature for the kehagias - sfetsos ( ks ) black hole in hoava - lifschitz gravity ( hl ) and finding no singularity in the curvature at a point where a direct calculation of the specific heat does indicate a singularity .
the legendre invariant metric suggested by quevedo _ et.al .
_ @xcite has proved more successful in capturing the phase structure of other black hole families than the ruppeiner metric , and in this note we calculate the quevedo metric and curvature for the ks black hole using both the entropy and mass as the thermodynamic potential . in the sequel
we first briefly define the action for hoava - lifschitz gravity and sketch the ks black hole solution .
we then describe the ruppeiner metric for the ks black holes , before moving on to the quevedo metric in both the entropy and mass representation .
an explicit expression @xmath5 for the ks black hole is presented , which is useful in calculating the metric and basic thermodynamic quantities . in conclusion ,
the general features of the scaling of the quevedo curvature and their origin are highlighted .
the suggestion by hoava @xcite that an anisotropic theory of gravity at a lifshitz point @xcite might offer a viable quantum field theory of gravity while still retaining the properties of einstein gravity in the ir has led to an explosion of recent work .
since the theory breaks general covariance to 3d spatial covariance plus time re - parametrization invariance it is most naturally couched in the ( 3 + 1 ) language of the adm @xcite formalism , where a general metric is written as @xmath6 the lapse and shift can then be used to construct the extrinsic curvature of the 3-space @xmath7 where the @xmath8 is the time derivative of the metric on the spatial slice .
the hoava - lifshitz action may then be written as @xmath9 with @xmath10 where @xmath11 , @xmath12 , @xmath13 , @xmath14 and @xmath15 are various constants and @xmath16 and @xmath17 are the three - dimensional ricci tensor and ricci scalar .
the cotton tensor for the three - geometry , which also appears , is defined as @xmath18 as it stands the generic ir vacuum of such a theory is anti - de sitter , but it is possible to deform the theory with an additional relevant operator @xmath19 , which allows a minkowski vacuum @xcite . using @xmath20 the ir limit of this augmented action matches the einstein - hilbert action @xmath21 in the limit @xmath22 and @xmath23 if @xmath24 the augmented action with @xmath25 considered in the limit @xmath22 is usually denoted deformed hl gravity " @xcite .
in @xcite kehagias and sfetsos showed that the deformed hl gravity at @xmath23 admits a schwarzschild - like black hole solution , where a metric ansatz @xmath26 leads to @xmath27 @xmath2 is an integration constant which is related to the mass of the black hole , as can be seen by noting that @xmath28 when @xmath29 which is the standard schwarzschild behaviour .
the ks black hole displays two event horizons at @xmath30 which is strikingly similar to the formula giving the event horizons in the reissner - nordstrm ( rn ) black hole in standard einstein gravity @xmath31 or that for the kerr black hole @xmath32 this has led to the suggestion that @xmath33 should be treated as a charge - like parameter when considering the thermodynamics of the ks black hole @xcite .
if one does this , the mass @xmath2 , hawking temperature @xmath4 and specific heat @xmath34 for the ks black hole may be written in a similar manner to those for the rn and kerr black holes giving @xcite @xmath35 the entropy may also be calculated and written in a beckenstein - smarr @xcite like manner , giving @xcite @xmath36 @xmath37 is a constant of integration which plays no role in the sequel .
it is natural to set @xmath38 to match up with the schwarzschild limit .
the singularity in the specific heat can be interpreted as signalling a phase transition at @xmath39 , i.e. @xmath40\ , r_{-}\ , \ , $ ] , for the ks black hole .
the ruppeiner metric components for the ks black hole were calculated by wei _
_ @xcite using equ.([sks ] ) @xmath41 the ricci scalar for the ruppeiner metric is then found to be @xmath42 ^ 2}\end{aligned}\ ] ] but it fails to show a singularity at the point , @xmath43 , where the specific heat has a singularity in equ.([cks ] ) , although the pre - factor of the log term in the denominator does vanish at this point .
in addition , it is neither zero nor singular in the extremal limit @xmath44 .
the ks black hole thus adds a further example to the ( long ) list of peculiarities which arise when information geometry , in the form of the ruppeiner or weinhold metric , is applied to the thermodynamics of black holes .
a possible resolution of such difficulties in general in the context of black hole thermodynamics was suggested by quevedo _
_ @xcite . they argued that an important feature for thermodynamic metrics was legendre invariance , which was not a property of either the ruppeiner or weinhold metrics .
they found consistent results for various black holes when using a legendre invariant metric definition , which agreed with direct calculations of phase transition points from the specific heat . in the next section
we apply the formalism to the ks black hole .
quevedo _ et al .
_ s starting point @xcite was the observation that standard thermodynamics was invariant with respect to legendre transformations , since one expects consistent results whatever starting potential one takes , and they coined the name geometrothermodynamics for a formalism which ensured this . their work was based on the use of contact geometry as a framework for thermodynamics , developed by hermann @xcite , mrugaa @xcite and others . for the geometrothermodynamics of black holes they considered a @xmath45 dimensional thermodynamic phase space @xmath46 with independent coordinates
@xmath47 , @xmath48 , where @xmath49 represents the thermodynamic potential , and @xmath3 and @xmath50 are the extensive and intensive thermodynamic variables , respectively .
this thermodynamic phase space was endowed with a gibbs one - form @xmath51 , @xmath52 , and the legendre transform invariant metric [ qg ] g & = & ( d- _ ab i^a d e^b)^2 + ( _ abe^a i^b)(_cd de^c di^d ) , + _ cd & = & diag(-1,1, ...
,1 ) , which was invariant with respect to @xmath53 , with @xmath54 .
the gibbs one - form satisfies the condition @xmath55 , making it a contact form and the triplet @xmath56 constitutes a riemannian contact manifold . the equilibrium space @xmath57 is then defined by @xmath58 , satisfying the condition @xmath59 .
this means that on @xmath60 the first law of thermodynamics holds , @xmath61 , and the equilibrium conditions @xmath62 give the @xmath50 in terms of the @xmath3 .
the induced thermodynamic metric on @xmath60 , which plays a similar role to the ruppeiner or weinhold metric and which we denote here as the quevedo metric , is given by g^q=(e^c ) ( _
ab^bc de^a de^d ) .
[ gq ] the choice of @xmath63 in equ.([qg ] ) rather than @xmath64 , which is also possible , prevents off diagonal terms @xmath65 , appearing which in turn plays a vital role in determining the singularities of the curvature . in the case of the ks black hole using the entropy
as the thermodynamic potential one considers the 5-dimensional thermodynamic phase space @xmath46 with coordinates @xmath66 . the fundamental one - form in this @xmath67representation is given by _ s = ds - dm +
d p , so defining the space of equilibrium states @xmath60 by @xmath68 , generates both the first law of thermodynamics of the ks black hole dm = t ds + v_p d p , [ first ] and the equilibrium conditions = , = - . from equ.([gq ] ) the quevedo metric in this case is g^q = ( m s_m + p s_p ) ( - s_mm d m^2 + s_pp dp^2 ) , [ grnq/ ] which may be written in components as @xmath69(r_+^2 - 5 r_+ r_- -2 r_-^2)}{(r_+ - r_-)^3},\nonumber\\ g^{q}_{12}&=&g^{q}_{21}= 0,\nonumber\\ g^{q}_{22}&= & - \frac{8 \pi^2 r_+ [ r_+ + 2 r_- + 4 r_- \ln ( r_+ ) ] a(r_+ , r_-)}{(r_+ - r_-)^3 } , \end{aligned}\ ] ] from which the ricci scalar may be calculated without further ado to give [ rksq ] r^q & = & + & & where a ( r_+ , r_- ) & = & r_+^3 + 10 r_+^2 r_- - 5 r_+ r_-^2 + 6 r_-^3 - 2 ( r_+ - r_-)^3 ( r_+ ) and @xmath70 is a long and not very illuminating expression which is neither zero nor divergent when @xmath71 and @xmath72 ( i.e. @xmath73 \ , r_-$ ] ) .
we give it in the appendix for completeness .
looking at the curvature @xmath74 we see that it diverges as @xmath75 , with @xmath76 , at the same point as the specific heat .
the specific heat vanishes in the extremal limit , @xmath77 since @xmath78 and @xmath79 in this limit .
this is also the case for the curvature in equ.([rksq ] ) calculated in the entropy representation , but this is not a generic feature , as we see in the next section by calculating the quevedo curvature using the mass as the thermodynamic potential .
there is a degree of arbitrariness in the definition of the metric @xmath80 on @xmath81 ( and hence @xmath82 ) , since various choices will allow for the desired legendre invariance .
different choices for the thermodynamic potential @xmath49 are also possible . in the preceding section we have taken @xmath83 , but
@xmath84 would have been _ a priori _ equally valid .
some of the properties such as the relation between the specific heat and curvature singularities of the quevedo metric are particularly apparent in this mass representation g^mq = ( s m_s + p m_p ) ( -m_ss d s^2 + m_pp dp^2 ) , [ gmq ] where , at the risk of overcomplicating the notation , we write the superscript @xmath85 to denote the use of the mass @xmath2 as the thermodynamic potential . in these variables the specific heat may be evaluated as @xmath86 and the presence of @xmath63 in the definition of @xmath87 ensures the absence of off - diagonal terms , @xmath88 and a @xmath89 factor in the resulting curvature . for the rn and kerr black holes
it is straightforward to express @xmath2 as a function of @xmath90 or @xmath91 using the beckenstein - smarr formulae [ msq ] m_rn(s , q ) = + m_kerr(s , j ) = ( ) ^1/2 .
it is more difficult to write @xmath5 explicitly for the ks black hole using equ.([sks ] ) , but this may still be done using the lambert w function , which is the solution of w(x ) ( w(x ) ) = x , to give m_ks ( s , p ) = , [ mw ] where we have assumed that @xmath92 .
choosing a non - zero constant simply shifts @xmath93 . the specific heat calculated from @xmath94 in these variables for the ks black hole is c = .
[ cksw ] rewriting this in terms of @xmath95 s reproduces ( as it should ) the expression for @xmath34 given in equ.([cks ] ) , which had been calculated previously @xcite using c = . the additional observation from solving equ.([mw ] ) for @xmath96 in terms of @xmath95 that @xmath96 is equal to @xmath97 is useful for rewriting equ.([cksw ] ) .
the curvature of the quevedo metric in the mass representation for the ks black hole is a rather cumbersome expression , but it takes the form r^mq = where @xmath98 are neither singular nor zero at the zeros of the other factor in the denominator .
this still shows clearly the correspondence between the specific heat and curvature singularities that is expected on general grounds , since substituting @xmath99 in the denominator recovers the same singular factor @xmath100 ^ 2 $ ] seen when the entropy is used as the thermodynamic potential .
the behaviour of the quevedo metric for the ks black hole is thus identical to that of the rn and kerr black holes : the location of the singularities of the curvature match those of the specific heat in both the entropy and mass representations .
with an explicit mass formula , equ.([mw ] ) , in hand it is also a straightforward matter to calculate the components of the weinhold metric for the ks black hole for comparison purposes with both the ruppeiner and quevedo metrics .
other black hole families display inconsistencies between the specific heat singularities and those of the ruppeiner and weinhold curvatures .
for example , with the rn black hole the ruppeiner geometry is flat , whereas the weinhold geometry is curved .
the ks black hole is no exception to this behaviour .
the metric components from @xmath101 are found to be @xmath102}{2 \pi r_+^{3/2 } ( r_+ + 2 r_-)^3},\nonumber\\ g^{w}_{22}&= & -\frac{1}{r_+ } + \frac{2 ( r_+^2 - 11 r_+ r_- - 2 r_-^2 ) \ln ( r_+ ) } { r_+ ( r_+ + 2 r_-)^2 } \nonumber \\ & + & \frac{8 r_- ( r_+^2 - 5 r_+ r_- - 2 r_-^2 ) \ln^2 ( r_+ ) } { r_+ ( r_+ + 2 r_-)^3 } , \end{aligned}\ ] ] which gives a curvature r^w & = & + & & where @xmath103 is another complicated expression with no interesting behaviour at @xmath72 or @xmath71 .
@xmath104 does _ not _ diverge at the same point as the specific heat , @xmath72 , but rather at the extremal limit .
interestingly , the factor in the denominator is identical to that appearing in the curvature of the ruppeiner metric , so the pre - factor of the @xmath105 disappears at that point . as for the ruppeiner metric
, the weinhold metric does not reproduce the singular behaviour of the ks black hole specific heat .
the ks black hole provides a further example of a system in which a particular choice of legendre invariant quevedo metric captures the phase structure in a manner which eludes the apparently physically well - motivated ruppeiner and weinhold metrics .
it remains a puzzle as to why this behaviour manifests itself in various black hole models and is not apparent when the ruppeiner and weinhold metrics are used in the description of other , less esoteric , statistical mechanical systems . in the latter case
, the use of a legendre invariant metric does not appear to be obligatory to find physically sensible results .
the relation between the singularities of the specific heat and the thermodynamic curvature calculated with this quevedo metric is consistent for the black holes and choices of thermodynamic potential discussed here .
if one accepts the premise that phase transitions appear as curvature singularities in the thermodynamic metric , none of the examples examined here and elsewhere using this particular definition have given rise to the sorts of inconsistencies which have dogged the application of ruppeiner and weinhold metrics in this field .
the relation between the specific heat and thermodynamic curvature scaling is different to that seen for a continuous transition and the ruppeiner or weinhold metrics in standard statistical mechanical systems . in such transitions
the ruppeiner curvature would be expected to diverge as the correlation volume , @xmath106 , where @xmath107 is some appropriate correlation length . if the standard scaling assumption @xmath108 holds , where @xmath109 at the critical point , @xmath110 .
if , in addition , hyperscaling @xmath111 is also valid , we find @xmath112 , which relates the singularity of the specific heat @xmath113 to the singularity of the curvature .
as we have seen , the specific heat for the black holes behaves as @xmath114 , but the thermodynamic curvature behaves as @xmath75 rather than @xmath115 .
the picture which emerges from the use of a quevedo metric to investigate the thermodynamics of the ks black hole is thus that , provided @xmath116 is treated as a charge , both the specific heat and the curvature from the quevedo metric possess singularities when @xmath72 .
the general behaviour is similar to other two parameter families such as the rn and kerr black holes , which is perhaps not so surprising given the similarities in the various thermodynamic formulae for the mass , charges and entropy .
this consistency might also be taken as further evidence that the choice of @xmath117 as a charge - like variable and the resulting expression for @xmath1 containing a logarithmic term in equ.([sks ] ) is the correct way to reconcile the integral and differential forms dm = t ds + v_p dp of the first law of thermodynamics in the ks black hole @xcite . as a final comment
, we emphasize that that the quevedo metric in the form used here recovers the singularities seen in the specific heat because certain choices ( in particular , no off - diagonal elements @xmath118 ) have been made
. it would be interesting to explore whether other choices might also be consistent , and what their physical motivation might be .
daj would like to thank h. quevedo for some useful clarifications .
the unilluminating factor @xmath119 in the numerator of @xmath120 in equ.([rksq ] ) : b ( , ) = 3 ^9 + 66 ^8 + 317 ^7 ^2 - 506 ^6 ^3 -3069 ^5 ^4 + - 5110 ^4 ^5 + 5221 ^3 ^6 + 114 ^2 ^7 + 284 ^8 + 88 ^9 + + - 4 ( ) [ 2 ^9 + 15 ^8 - 78 ^7 ^2 -141 ^6 ^3 -558 ^5 ^4 + + 3507 ^4 ^5 -3402 ^3 ^6 - 1293 ^2 ^7 -648 ^8 + 4 ^9 ] + + + 8 ( ) ^2 ( - ) [ ^8 - 3 ^7 + 4 ^6 ^2 -306 ^5 ^3 + 773 ^4 ^4 + -1227 ^3 ^5 -694 ^2 ^6 -284 ^7 + 8 ^8 ] + + - 32 ( ) ^3 ( - ) ^2 [ ^5 -12 ^4 + 38 ^3 ^2 -64 ^2 ^3 + - 51 ^4 -20 ^5 ] h. janyszek and r. mrugaa , phys
. rev . * a 39 * ( 1989 ) 6515 ; + h. janyszek , rep . math
* 24 * ( 1986 ) 1 ; _ ibid _ 11 .
d. brody and n. rivier , phys .
e 51 * ( 1995 ) 1006 ; + d. brody and a. ritz , nucl .
* b 522 * ( 1998 ) 588 ; + e.j .
brody , phys .
* 58 * ( 1987 ) 179 ; + d. brody and l. hughston , proc .
london * a 455 * ( 1999 ) 1683 . | the application of information geometric ideas to statistical mechanics using a metric on the space of states , pioneered by ruppeiner and weinhold , has proved to be a useful alternative approach to characterizing phase transitions .
some puzzling anomalies become apparent , however , when these methods are applied to the study of black hole thermodynamics . a possible resolution was suggested by quevedo _
et al . _ who emphasized the importance of legendre invariance in thermodynamic metrics .
they found physically consistent results for various black holes when using a legendre invariant metric , which agreed with a direct determination of the properties of phase transitions from the specific heat .
recently , information geometric methods have been employed by wei _
et al .
_ to study the kehagias - sfetsos ( ks ) black hole in hoava - lifshitz gravity .
the formalism suggests that a coupling parameter in this theory plays a role analogous to the charge in reissner - nordstrm ( rn ) black holes or angular momentum in the kerr black hole and calculation of the specific heat shows a singularity which may be interpreted as a phase transition .
when the curvature of the ruppeiner metric is calculated for such a theory it does not , however , show a singularity at the phase transition point .
we show that the curvature of a particular legendre invariant ( `` quevedo '' ) metric for the ks black hole _ is _ singular at the phase transition point .
we contrast the results for the ruppeiner , weinhold and quevedo metrics and in the latter case investigate the consistency of taking either the entropy or mass as the thermodynamic potential . |
calculation of the cross section of central exclusive diffraction , @xmath4 ( @xmath5 = dijet , heavy quarkonium , higgs boson , _
etc_. ) presents a major challenge for strong interaction physics .
it involves treating the hard dynamics in the elementary @xmath2 subprocess , and calculating the probability that no other interactions leading to hadron production occur during the @xmath1 collision .
the latter determines the suppression of diffractive relative to non - diffractive events with the same hard process , referred to as the rapidity gap survival ( rgs ) probability . in this article
we summarize our understanding of the dynamical mechanisms determining the rgs probability , their phenomenological description , and the uncertainties in present numerical predictions .
rgs in central exclusive diffraction has extensively been discussed in an approach where soft interactions are modeled by eikonalized pomeron exchange ; see ref .
@xcite for a summary .
more recently a partonic description was proposed , which allows for a model independent formulation of the interplay of hard and soft interactions and reveals the essential role of the `` transverse geometry '' of the @xmath1 collision @xcite . in the mean field approximation , where hard and soft interactions are considered as independent aside from their common dependence on the impact parameter
, we derived a simple `` factorized '' expression for the rgs probability , using closure of the partonic states to take into account inelastic diffractive intermediate states .
the resulting rgs probability is smaller than in the models of refs .
@xcite without inelastic diffraction , but comparable to the some of the versions of those models with multichannel diffraction .
our partonic description also permits us to go beyond the mean field approximation and incorporate various types of correlations between the hard scattering process and spectator interactions .
here we discuss two such effects : ( a ) quantum fluctuations of the partonic configurations of the colliding protons , which somewhat reduce the survival probabilities at rhic and tevatron energies ; ( b ) absorptive interactions of high - virtuality spectator partons ( @xmath6 ) associated with the hard scattering process , related to the onset of the black disk regime ( bdr ) in hard interactions at lhc energies ; this new effect substantially reduces the rgs probability compared to previously published estimates .
a simple picture of rgs is obtained in the impact parameter representation .
on one hand , to produce the heavy system @xmath5 two hard gluons from each of the two protons need to collide in the same space time point ( actually , an area of transverse size @xmath7 in the hard process ) ; because such gluons are concentrated around the transverse centers of the protons this is most likely when the protons collide at a small impact parameters , @xmath8 . on the other hand ,
soft inelastic spectator interactions are strongest at small @xmath9 and would favor collisions at @xmath10 for diffractive scattering .
these different preferences limit diffraction to an intermediate range of impact parameters and ensure that its cross section is substantially suppressed compared to non diffractive scattering .
more precisely , the rgs probability is given by @xcite @xmath11 here @xmath12 is the probability for two gluons to collide at the same transverse point as a function of the @xmath1 impact parameter , given by the convolution of the transverse spatial distributions of the gluons in the colliding protons , normalized to @xmath13 ( see fig .
[ fig : rgs]a ) .
the factor @xmath14 is the probability for the two protons not to interact inelastically in a collision at the given impact parameter , calculable in terms of the profile function of the @xmath1 elastic amplitude , @xmath15 .
figure [ fig : rgs]b shows the @xmath9dependence of the two factors as well as their product , illustrating the interplay described above . while we have motivated eq .
( [ survb ] ) by probabilistic arguments , it actually can be derived ( as well as the expression for the differential cross section ) in the partonic description of ref
. @xcite within the mean field approximation , where one assumes no correlation between the presence of the gluons involved in the hard interaction ( with the particular @xmath16 ) and the strength of the soft spectator interactions . in this approximation
one can use closure to sum over the different diffractive intermediate states , and thus effectively include the contribution of inelastic diffraction . for generic @xmath1 diffraction and @xmath17 for the gpd ) ; see section iv c of ref .
@xcite . ] the numerical values of the rgs probability obtained from eq .
( [ survb ] ) are of the order @xmath18 for @xmath19 and @xmath20 ; see ref .
@xcite for details .
[ cols="<,^ , < " , ] it is worthwhile to discuss the uncertainty in the numerical predictions for @xmath21 in the mean field approximation , eq .
( [ survb ] ) , resulting from our imperfect knowledge of the functions in the integrand .
we first consider the transverse spatial distribution of gluons entering in @xmath22 .
the latter is obtained as the fourier transform of the @xmath23dependence ( more precisely , transverse momentum dependence ) of the gluon generalized parton distribution ( gpd ) measured in hard exclusive vector meson production .
extensive studies at hera have shown that exclusive @xmath24 photoproduction , @xmath25 , provides an effective means for probing the @xmath23dependence of the gluon gpd at small and intermediate @xmath16 ( a small correction for the finite transverse size of the @xmath24 is applied ) @xcite .
figure [ fig : bpsi ] summarizes the results for the exponential @xmath23slope of this process , @xmath26 , from hera h1 @xcite and zeus @xcite and the fnal e401/e458 experiment @xcite , as well as fits to the @xmath16dependence of the h1 and zeus results of the form ( here @xmath27 ) @xmath28 there is a systematic difference between the h1 and zeus results due to different analysis methods @xcite ; however , the fits to both sets agree well with the fnal point when extrapolated to larger @xmath16 . in diffractive production of a system with @xmath19 at @xmath29 at zero rapidity the gluons coupling to the heavy system
@xmath5 have momentum fractions @xmath30 . assuming exponential @xmath23dependence of the gluon gpd , we can estimate the uncertainty in the transverse spatial distribution of gluons at such @xmath16 by evaluating the fits to the hera data within the error bands quoted for @xmath31 and @xmath32 @xcite .
we find a 15 - 20% uncertainty of @xmath26 at @xmath33 in this way , translating into a 2030% uncertainty in the mean field rgs probability , eq .
( [ survb ] ) .
we note that there is at least a comparable uncertainty in @xmath21 from the uncertainty of the shape of the @xmath23dependence ; this is seen from fig .
10 of ref .
@xcite , where the exponential is compared with a theoretically motivated dipole form which also describes the fnal data .
altogether , we estimate that our imperfect knowledge of the spatial distribution of gluons results in an uncertainty of the mean field result for @xmath21 by a factor @xmath34 .
dedicated analysis of the remaining hera exclusive data , and particularly precision measurements with a future electron ion collider ( eic ) , could substantially improve our knowledge of the transverse spatial distribution of gluons .
we now turn to the uncertainty in @xmath21 arising from the @xmath1 elastic amplitude , @xmath35 .
most phenomenological analyses of @xmath1 elastic and total cross section data find that for tev energies @xmath36 at @xmath37 , corresponding to near unit probability of inelastic interactions at small impact parameters ( bdr ) .
this is supported by theoretical studies in the qcd dipole model , which show that the large@xmath16 partons with virtualities of up to several @xmath38 experience `` black '' interactions with the small@xmath16 gluon field in the other proton when passing through the other proton at transverse distances @xmath39 , and receive transverse momenta @xmath40 ( see ref .
@xcite for a summary ) . at @xmath1 impact parameter @xmath37
the chance that none of the leading partons in the protons receive such a kick is extremely small , implying that @xmath41 @xcite . for the rgs probability in the mean field approximation , eq .
( [ survb ] ) , the fact that @xmath42 is small at @xmath37 is essential , as this eliminates the contribution from small @xmath9 in the integral ( see fig .
[ fig : rgs]b ) and stabilizes the numerical predictions
. however , present theoretical arguments and data analysis can not exclude a small non - zero value of @xmath43 at @xmath37 ; a recent analysis finds @xmath44 @xcite . to investigate the potential implications for the rgs probability , we evaluate eq .
( [ survb ] ) with the gaussian parametrization of @xmath35 of ref .
@xcite , eq .
( 12 ) , but with @xmath45 .
we find that a value of @xmath46 , corresponding to @xmath47 , increases the mean field result for @xmath21 by a factor @xmath48 , indicating significant uncertainty of the mean field result .
however , as explained in sec .
[ sec : bdr_hard ] below , hard spectator interactions associated with the @xmath2 process lead to an additional suppression of diffraction at small @xmath9 ( not contained in the soft rgs probability ) , which mitigates the impact of this uncertainty on the overall diffractive cross section .
corrections to the mean field picture of rgs arise from fluctuations of the interacting configurations in the colliding protons .
this concept is known well in soft diffraction , where fluctuations of the strength of interaction between the colliding hadrons give rise to inelastic diffraction . in hard diffraction ,
one expects that also the gluon density fluctuates ; _
e.g. _ because the color fields are screened in configurations of small size @xcite .
in fact , the variance of the gluon density fluctuations can be directly related to the ratio of inelastic and elastic diffraction in processes such as @xmath49 , @xmath50 the hera data are consistent with the dynamical model estimate of @xmath51 for @xmath52 and @xmath53 @xcite ; unfortunately , the limited @xmath54 range and the lack of dedicated studies do not allow for a more precise extraction of this fundamental quantity . in central exclusive diffraction , correlated fluctuations of the soft interaction strength and the gluon density lower the rgs probability , because small - size configurations which experience little absorption have a lower gluon density .
this effect can be modeled by a generalization of the mean field expression ( [ survb ] ) , in which both the gluon gpds in @xmath55 and the profile function fluctuate as a function of an external parameter controlling the overall size of the configurations @xcite .
numerical studies find a reduction of the rgs probability by a factor @xmath56 for a system with mass @xmath57 produced at zero rapidity at @xmath58 . the dynamical model used in this estimate does not include fluctuations of the gluon density at larger @xmath59 , which could increase the suppression .
we emphasize again that inelastic diffraction _
per se _ is included in the partonic approach of ref .
@xcite through the closure of partonic states .
the effect discussed in this section is specifically related to correlations between the fluctuations of the parton densities and the soft interaction strength ; in the limit of zero correlations ( independent fluctuations ) we recover the mean field result described above @xcite .
substantial changes in the mechanism of diffractive scattering are brought about by the onset of the bdr in hard interactions at lhc energies , where even highly virtual partons ( @xmath60 ) with @xmath61 experience `` black '' interactions with the small@xmath16 gluons in the other proton .
this new effect modifies the amplitude of central exclusive diffraction in several ways : ( a ) absorption of the `` parent '' partons of the gluons attached to the high mass system ; ( b ) absorption of the hard gluons attached to the high mass system ; ( c ) absorption due to local interactions within the partonic ladder .
such absorptive hard interactions cause additional suppression of diffractive scattering , not included in the traditional soft interaction rgs probability @xcite .
because of the generic nature of `` black '' interactions , we can estimate this effect by a certain modification of the mean field picture in the impact parameter representation . here
we focus on mechanism ( a ) and show that it causes substantial suppression ; the other mechanisms may result in further suppression . .
( b ) absorptive interaction of the hard spectator with small@xmath16 gluons in the other proton.,scaledwidth=98.0% ] according to ref .
@xcite ( and references therein ) the dominant contribution to the hard amplitude of higgs production at the lhc ( @xmath62 ) originates from gluons with transverse momenta of the order @xmath63 .
such gluons are typically generated by dglap evolution starting from the initial scale , @xmath64 , in which spectator partons , mostly gluons , are emitted ( see fig.[hardabs]a ) . in the leading log approximation @xmath65 , and
thus the transverse distance between the active and spectator parton is @xmath66 , amounting to short range correlations between partons . if the interactions of the spectator parton with the small@xmath16 gluons in the other proton become significant ( see fig.[hardabs]b ) , the basic assumption of the mean field approximation
that the spectator interactions are independent of the hard process is violated , and the interactions of that parton need to be treated separately . indeed , studies within the qcd dipole model show that at the lhc energy spectator gluons with @xmath67 and @xmath68 `` see '' gluons with momentum fractions @xmath69 in the other proton , and are absorbed with near unit probability if their impact parameters with the other proton are less than @xmath70 @xcite . ) dipoles is larger than that of the quark antiquark ( @xmath71 ) dipoles in @xmath72 scattering @xcite by a factor @xmath73 . a summary plot of the profile function for gluon proton scattering is given in fig .
13 of ref .
@xcite ( right @xmath74axis ) .
note that @xmath75 already corresponds to a significant absorption probability of @xmath76 . ] for @xmath1 impact parameters @xmath77 about @xmath78 of the strength in @xmath22 comes from parton
proton impact parameters @xmath79 ( _ cf_. fig .
[ fig : rgs]a ) , so that this effect practically eliminates diffraction at @xmath77 . since @xmath77 accounts for 2/3 of the cross section ( see fig .
[ fig : rgs]b ) , and the remaining contributions at @xmath80 are also reduced by absorption , we estimate that absorptive interactions of hard spectators in the bdr reduce the rgs probability at lhc to about 20% of its mean field value .
much less suppression is expected at the tevatron energy , where hard spectator interactions only marginally reach the bdr . in the above argument
one must also allow for the possibility of trajectories with no gluon emission , which correspond to the sudakov form factor
suppressed @xmath81term in the evolution kernel . while such trajectories are not affected by absorption ,
their contributions are small both because of the sudakov suppression , and because they effectively probe the gluon density at a low scale , @xmath82 , where evolution induced correlations between partons can be neglected .
we estimate that the contribution of such trajectories to the cross section is suppressed compared to those with emissions by a factor @xmath83 ^ 2 \sim 1/10 $ ] , where @xmath84 $ ] is the square of the sudakov form factor , and @xmath85 .
their net contribution is thus comparable to that of the trajectories with emissions , because the latter are strongly suppressed by the absorption effect described above . combining the two
, we obtain an overall suppression by a factor of the order @xmath86 .
more accurate estimates would need to take into account fluctuations in the number of emissions ; in particular , trajectories on which only one of the partons did not emit gluons are suppressed only by @xmath87 and may make significant contributions .
the absorptive hard spectator interactions described here `` push '' diffractive @xmath1 scattering to even larger impact parameters than would be allowed by the soft spectator interactions included in the mean field rgs probability , eq .
( [ survb ] ) ( except for the sudakov suppressed contribution ) .
one interesting consequence of this is that it makes the uncertainty in the mean field prediction arising from @xmath88 ( see sec .
[ sec : mean ] ) largely irrelevant , as the region of small impact parameters is now practically eliminated by the hard spectator interactions .
another consequence is that the final state proton transverse momentum distribution is shifted to to smaller values ; this could in principle be observed in @xmath89dependent measurements of diffraction .
we note that the estimates of hard spectator interactions reported here are based on the assumption that dglap evolution reasonably well describes the gluon density down to @xmath90 ; the details ( but not the basic picture ) may change if small@xmath16 resummation corrections were to significantly modify the gluon density at such values of @xmath16 ( see ref .
@xcite and references therein ) .
the approach to the bdr in the interaction of hard spectator partons , caused by the increase of the gluon density at small @xmath16 , has profound implications for central exclusive diffraction at lhc : _ no saturation without disintegration ! _
the rgs probability is likely to be much smaller ( by a factor of @xmath91 or less ) than predicted by the mean field approximation or corresponding models which neglect correlations of partons in the transverse plane .
diffractive scattering is relegated either to very large impact parameters ( @xmath92 ) or to sudakov suppressed trajectories without gluon radiation .
we estimate that the overall rgs probability at lhc is @xmath93 .
extrapolation of the tevatron results may be misleading because interactions of hard spectators are generally far from `` black '' at that energy .
the new effects described here call for detailed mc based studies of possible histories of the hard scattering process and their associated spectator interactions . | we summarize our understanding of the dynamical mechanisms governing rapidity gap survival in central exclusive diffraction , @xmath0 , and discuss the uncertainties in present estimates of the survival probability .
the main suppression of diffractive scattering is due to inelastic soft spectator interactions at small @xmath1 impact parameters and can be described in a mean field approximation ( independent hard and soft interactions ) .
moderate extra suppression results from fluctuations of the partonic configurations of the colliding protons . at lhc energies absorptive interactions of hard spectator partons associated with the @xmath2 process reach the black disk regime and cause substantial additional suppression , pushing the survival probability below @xmath3 . |
the nasa deep space network ( dsn ) is a world - wide facility for communicating with deep space missions .
transient observatory ( dto ) is a signal processing facility that can monitor up to four dsn telemetry bands for astronomically interesting signals .
since telemetry signals occupy only a small fraction of the entire band , the rest of the band can be used for commensal scientific investigations .
the signal processing requirements may be quite different for these projects .
dto supports these by using different firmware on the same platform . currently , the primary scientific objective is monitoring mars for electrostatic discharges .
such bursts , reported by @xcite , occurred during a convective dust storm .
since the dsn is receiving data from spacecraft at mars almost continuously , a statistically significant database can reveal the conditions which give rise to such events .
other planets can also be monitored for transient phenomena .
additional investigations include searching for astronomical transients such as fast radio burts ( frbs ) , pulsars and signals resulting from extraterrestrial technologies .
there is one 70 m diameter antenna in goldstone , california , with others near canberra , australia , and madrid , spain , roughly equidistant in longitude to enable continuous communication with robotic missions exploring the solar system .
each site also has at least three 34-m antennas .
table [ tab : bands ] @cccc@ band & frequencies & bandwidth & polarization + & ( ghz ) & ( mhz ) + s & 2.265 - 2.305 & 40 & lcp @xmath0 rcp + x & 8.25 - 8.65 & 400 & lcp & rcp + ka & 31.85 - 32.25 & 400 & lcp & rcp + the polarization for s - band is configured at the beginning of a tracking session .
it is normally rcp .
only some 34-m antennas have s - band .
[ tab : bands ] shows the bands used for deep space to earth communications .
the initial signal processing is done with two roach1 boards , each handling up to two 640mhz - wide baseband signals .
each roach1 has a 10 gb ethernet ( 10 gbe ) interface with a gpu - equipped debian linux workstation for additional processing .
the monitoring is done commensally with reception of deep space mission signals .
since deep space missions move slowly across the sky , the time for acquiring data from a given direction can be quite long , though the direction is not under the investigator s control . at minimum elongation
, venus moves about one beamwidth per hour at x - band for a 70 m antenna or at ka - band for a 34 m antenna .
for mars the minimum dwell time is about 100 min . for jupiter and saturn
it is about 5@xmath1 and 11 hr respectively .
if , one the other hand , one wishes to survey more sky quickly , for a 34 m antenna that is not moving , the dwell time for a position on the sky is @xmath2s at s - band , @xmath3s at x - band and 3.8s at ka - band . a non - tracking 34-m antenna can cover 1% of the sky in 17 earth rotations at s - band , 59 at x and 227 at ka .
dto can also be used during tracking sessions assigned to observations of radio astronomical sources .
the facility will be available to the astronomical community through a peer - reviewed proposal process .
figure [ fig : overview ] shows an overview of the signal processing facility . a commercial @xmath4 intermediate frequency ( if ) matrix switch made by jfw industries , inc .
, directs up to four station ifs to the inputs of two roach1 signal acquisition and processing units .
two high - speed , multi - core , gpu - equipped workstations accept processed data from each roach over a 10 gb ethernet ( 10gbe ) port .
the powerpcs on the roach boards boot their kernel images from the master controller dto .
dto also has the file systems of the powerpcs which control the operation of a xilinx xc5vsx95 t field programmable gate array ( fpga ) .
the bit - files which configure the roach fpgas are loaded by the powerpcs .
so , in effect , dto provides the firmware for the fpgas , allowing rapid switching of firmware under computer control .
the sample clock generator ( samgen ) subsystem provides the clock signals for the analog - to - digital convertors ( adc ) , optional synchronization pulses and if amplification and low - pass filtering . figure [ fig : samgen ] has a schematic diagram .
samgen has a valon dual synthesizer so the two roach board adcs can be independently clocked .
it also distributes 1pps synchronization pulses to the kat adcs .
in addition to the controller dto and the two ppcs , there are two high - performance computers .
each post - processing ppc has two intel westmere 2.66ghz x5650 processors with six cores , allowing it to run twelve threads .
it has 96 gb of ram , two gtx 580 gpu processors each with 512 cores , and 48 tb of fast disk .
32 t are available for data storage .
all hosts have the debian linux operating system .
on 2006 june 8 deep space station 13 ( dss-13 ) detected non - thermal bursty emissions from mars @xcite that showed resonances similar to schumann resonances on earth @xcite .
transient non - thermal radio bursts can be distinguished from gaussian thermal noise by computing the kurtosis ( normalized fourth moment ) of the signal voltage @xcite .
for such studies we have designed specifically for dto a 1024-ch firmware spectrometer that computes kurtosis as well as power .
the design is shown in figure [ fig : kurtspec ] .
schematic diagram of the kurtosis spectrometer firmware , including all the registers and data stores.,width=672 ] for mars , the integration time for these calculations , a parameter which can be adjusted , is 1 ms .
this allows a signal kurtosis spectrogram to be examined for the presence of modulation by very low frequency waves propagating between the surface and ionosphere of mars .
the dsn has telemetry with at least one spacecraft at mars for an average of 20 hrs every day .
one of dto s main goals is to monitor mars diligently for further evidence for electrostatic discharges
. it would be unlikely to detect radio spectral line emission from an arbitrary direction in the sky but planetary atmospheres can produce such emission from very low density upper atmospheres .
one of the available firmwares is a 32k - channel spectrometer .
it was developed for the tidbinbilla agn maser survey ( tams ) project led by harvard smithsonian astrophysical observatory @xcite and is part of the 17 - 27 ghz receiver system on dss-43 @xcite . with a 640 mhz bandwidth
this gives a spectral resolution of 20 khz .
the doppler velocity resolution is 2.5 km s@xmath5 at s - band , 0.7 at x - band and 0.2 km s@xmath5 at ka - band .
this firmware is most likely to be used for antenna time assigned to astronomical research
. it would also be useful for monitoring radio frequency interference ( rfi ) and spacecraft transmitters which have lost lock and drifted in frequency .
extraterrestrial civilizations that are located close to the sun s ecliptic plane could be aware of earth and its life - supporting potential from earth s transits across the sun .
such a civilization might direct a beacon towards earth to initiate possible contact .
an interesting strategy is to conduct searches for extraterrestrial civilizations ( seti ) along the ecliptic plane @xcite .
the dsn , in communication with spacecraft at or travelling towards other planets , has many antennas pointed close to the ecliptic .
a commensal seti search may have enhanced chances of success compared to similar searches at radio observatories @xcite .
there are 82 k and g stars within one kpc of earth within this zone @xcite , so that over a time comparable to the orbital periods of the planets , they would be within the beam of antennas communicating with deep space missions .
amplitude modulated signals transmitted by putative extraterrestrial civilizations are expected to be most easily detected by the carrier tone , which is expected to be extremely narrow .
an extremely high resolution spectrometer design , serendip vi , will be adapted from roach2 .
the firmware performs 4k channelization .
the samples from each channel of the firmware output are further channelized in software using gpus . a 128k pfb on each channel from the firmware spectrometer
will yield a spectral resolution of 1.2 hz .
serendip vi has been implemented in setiburst , a multi - purpose signal processor @xcite similar in concept to dto .
the high resolution of the final spectra allows masking of spacecraft telemetry and known interference signals .
since the amount of data is still huge , additional back - end processing will decimate the data .
for example , the jpl `` setispec '' used in the gavrt seti project @xcite , reports for each of 4096 coarse channels , the number of and power detected in the strongest high - resolution sub - channel .
the monitor and control ( m&c ) software has a server / client architecture .
all control software is written in python . inter - process communication ( ipc )
is managed with pyro .
overview of the dto software.,width=470 ] figure [ fig : sw_oview ] gives a schematic overview of the software .
the * scheduler * processes the dsn 7-day schedule for the goldstone antennas and selects antennas and frequency bands according to a rule set determined by the science team .
the * supervisor * is a client of the hardware servers and monitors and controls all aspects of dto the state of the if switches , the frequency and power of the sample clock generators , the gain of the katadc rf sections , the firmware loaded into the fpgas , etc .
it also collects metadata from the station s monitor data server using the dsn s ipc protocol .
the can be overseen by a human through the supervisor client .
this has a set of classes and functions through which an expert user can communicate with the supervisor through a python command line .
a graphical user interface ( gui ) can invoke the same classes and functions to provide a more convenient overview .
the * roach power pcs * control most of the roach operation .
the ppc filesystems are on dto .
the supervisor loads the appropriate firmware for a session into the roachs .
the user s client software works through a tunnel @xcite which is created by a user with the requisite authority ( username and two - factor passcode ) to log in to the firewall s gateway .
only one log - in is needed for a session .
data generated by the firmware streams through 10gbe connections to the signal processing hosts . because of the generally high data rate involved , this software is normally written multi - threaded c. python s libraries for gpu support are being investigated . the ouput is written to disk for further real - time analysis . in the final software ,
the data are stored temporarily in a circular buffer .
these data are examined for `` events '' .
this software is the most complex as it must separate real from false events , generate alerts for events requiring prompt attention and store data which events of interest .
dto is awaiting shipment to the goldstone deep space communications complex ( gdscc ) .
the hardware , firmware and m&c software for the initial science program has been verified in the laboratory at jpl .
the location and access to power and signals at goldstone have been assigned .
paperwork required for installation in the configuration - controlled dsn operations environment has been completed .
dto will become operational at goldstone in the late summer of 2016 .
dto will be operated as an open facility .
the jpl interplanetary network directorate ( ind ) which manages the dsn will accept proposals from outside investigators who bring their own firmware to conduct research with different goals .
because of the flexibility built into the casper development ecosystem , dto should not be considered limited to the applications described here , but rather a prototype for general purpose commensal science signal processing . for example , the high resolution spectrometer design was recently implemented as four spectrometers on a single roach2 . additional modifications , such as extending the kurtosis based designs for flexible integration times , and development of higher order ( 6th , 8th moment ) statistics are also ongoing .
the instituto nacional de tcnica aeroespacial ( inta ) and ingeniera de sistemas para la defensa de espaa ( isdefe ) have made available four roach1 boards together with a controller host to the radio astronomy department of the madrid deep space communications complex ( mdscc ) .
inta and jpl will collaborate in assembling a second dto for mdscc which will be operated by inta scientists .
a fifth roach1 board will be used to provide a spectrometer for the partner educational antenna ( proyecto acadmico del radiotelescopio de nasa en robledo ) , a project also managed by inta .
the deep space network is operated by the california institute of technology jet propulsion laboratory for the national aeronautics and space administration ( nasa ) .
this work was funded by the science and technology investment program of the offices of the jpl chief scientist and chief technologist .
we are endebted to jonathan kocz for completing the kurtosis firmware s 10 gbe output capability .
we thank chris ruf and nilton renno for insightful discussions on kurtosis and the science of electrostatic discharges in the martian atmosphere .
the advice and support of the jpl interplanetary network directorate chief scientist , joseph lazio , is gratefully acknowledged .
chuck naudet leads the dsn science automation task which develops software that will enhance dto operation .
dong shin assisted in assembling dto . | the dsn transient observatory ( dto ) is a signal processing facility that can monitor up to four dsn downlink bands for astronomically interesting signals .
the monitoring is done commensally with reception of deep space mission telemetry .
the initial signal processing is done with two casper roach1 boards , each handling one or two baseband signals .
each roach1 has a 10 gbe interface with a gpu - equipped debian linux workstation for additional processing .
the initial science programs include monitoring mars for electrostatic discharges , radio spectral lines , searches for fast radio bursts and pulsars and seti
. the facility will be available to the scientific community through a peer review process . ; ; ; |
at an early stage in their evolution stars eject material in the form of outflows .
in fact , the first signs of an outflow are coupled with infall motion and therefore with the first stages of star formation ( e.g. bachiller 1996 ; richer et al .
once the protostar is formed , outflows are the main means of removing the material left over from the collapse of the cloud .
small molecular clumps ( @xmath2 0.1 pc ) are detected in association with several outflows ; it is generally believed that the clumps are generated by episodic mass loss of the forming object . to date , there is no detailed understanding of the role of the clumpiness in outflows .
in fact , two main kinds of clumps have been observed : ( i ) high - velocity clumps or the so - called molecular bullets which are well defined entities travelling at velocities larger than 100 km s@xmath3 .
the prototype is l1448 ( bachiller , martn - pintado & fuente 1991 ) , where bullets appear in pairs with the members of a pair being symmetric in both position and velocity with respect to the star .
these bullets are most likely associated with mini - bow shocks formed by the outflow propagation ( e.g. dutrey , guilloteau & bachiller 1997 ) .
the molecular lines emitted by this kind of clump are relatively weak , so their chemical composition remains unknown .
( ii ) a second kind of clump , observed along a few outflows associated with low- and intermediate - mass stars ; these outflows stand out because of their association with chemically rich clumps at definitely lower velocity , such as l1157 , bhr71 , and cb3 ( bourke et al .
1997 ; codella & bachiller 1999 ; bachiller et al .
the origin of these chemically rich clumps is not yet clear . in this paper
, we investigate the origin and nature of the second kind of clumps by the use of a chemical model that simulates the clump formation and its subsequent interaction with the outflow .
we consider here two main scenarios : ( 1 ) pre - existing clumps , affected by the outflow , and ( 2 ) newly - formed clumps , created by the outflow . _
note that we will use the word pre - existing to indicate a density structure formed before the advent of the outflow .
_ our definition does not imply that the observed abundances ( e.g codella & bachiller 1999 ; bachiller et al .
2001 ) are pre - existing to the outflow . in
what follows , we introduce our two scenarios for the formation of the clumps . *
the clumps are pre - existing to the outflow ; for example they may be either remnant material of the collapsing parent cloud or completely independent of the star formation process , but present homogeneously in the dark molecular cloud ( e.g. falle & hartquist 2002 ; morata et al .
2003 ; garrod et al .
2003 ) . in this scenario
, we observe these clumps in association with molecular outflows because , as the outflows travel through the molecular cloud , they interact with the clumps and shock them .
if this is the case , then the clumps may not be an indication of the episodic nature of outflows . *
the cloud material is homogeneous and low in density but as an episodic outflow forms , its interaction with the surrounding homogeneous material will lead to a compression ( increase in density ) and increase in temperature ( with subsequent evaporation of the grains mantles ) in localized regions _ only _ , hence the observed clumpiness ( e.g arce & goodman 2001 , 2002 , and reference therein ) . in this scenario ,
the clumps are a direct manifestation of the episodic nature of outflows .
note that these scenarios do not necessarily exclude each other : in fact , the morphology of the regions where outflows and jets propagate have a very complicated geometry and structure by their very nature ( e.g hester et al . 1998 ) .
the main aim of this study is to investigate the nature of the chemically rich clumps by modelling their chemical evolution and by comparing the models with observations . in particular , we attempt to identify observable species that can be used as discriminants between the two scenarios . in this paper
, we will focus our attention on the clumps located in the intermediate - mass star forming region cb3 ( codella & bachiller 1999 ) .
in addition , a low mass case , represented by the l1157 outflow , will be briefly discussed .
our model is described in section 2 , and our results are presented , discussed and compared with observations in section 3 . a brief conclusion is given in section 4 .
the aim of the modelling is to explore the chemistry in the different scenarios of clump formation described above .
the basic chemical model we adopt is a modification of the time - dependent model employed in viti & williams ( 1999 ) and viti et al .
the chemical network is taken from the umist database ( millar et al .
1997 ; le teuff et al .
we follow the chemical evolution of 221 species involved in 3194 gas - phase and grain reactions .
our model is two - phase calculation .
phase i represents the formation of the pre - existing clumps , scenario 1 , or of the homogeneous dark cloud , scenario 2 , from diffuse gas , while phase ii represents the effect on the gas and dust of the outflow .
figure [ fg : scheme ] shows a schematic of how the two scenarios are treated in the model . in phase
i , we allow the diffuse gas to collapse gravitationally within the molecular cloud . during the collapse phase , gas - phase chemistry and freeze - out on to dust grains with subsequent processing
are assumed to occur .
the initial density of the gas is taken to be 100 @xmath4 , and the final density is treated as a free parameter ( see later ) . during phase i , the temperature is kept constant at 10 k. in phase ii we simulate the presence of the outflow ( see sections 2.12.3 for a detailed description of each grid of models ) by an increase in temperature and an assumed subsequent thermal or ( non - dissociative ) shock - induced evaporation of the grain mantles .
sputtering of the core of the grains and fast , dissociative shocks are not included .
phase ii was halted at 10@xmath5 yr .
in phase ii , scenario 1 , we have investigated both a uniform and homogeneous clump ( the set of physical parameters for which we have explored the chemistry is here called grid a ) , and clumps with a density structure ( grids b and c ) .
we have also considered the possibilities that after the initial increase of the temperature ( to @xmath2 100 k ) , the temperature remains constant ( grid c ) or enters a shocked phase to a higher temperature , followed by cooling ( grid b ) .
see 2.12.3 for more details . in scenario 2 the outflow impacts on a dark molecular cloud , of density @xmath2 10@xmath6 @xmath4 , and induces the formation of clumps . in this case
we use phase i of the model to simulate the formation of the dark cloud , and phase ii to simulate the formation of the clumps due to the impact of the outflow . during phase ii ,
the gas and dust temperatures increase as the clump forms , and the grains evaporate . here
too we investigated the possibility that after the initial increase to 100 k , the temperature enters a shocked phase to a higher temperature and subsequent cooling .
this scenario will be modelled in a range of physical parameters here labelled grid d. within each grid we have explored a reasonably large parameter space .
to impose some constraints to the initial parameters , we adopt the physical parameters derived by codella & bachiller ( 1999 ) for the clumps observed along cb3 , namely : a size for the final clump of @xmath2 0.12 pc , a final density of about 10@xmath510@xmath7 @xmath4 , a final temperature of 100 k , and an age of 10@xmath5 yrs .
table [ tb : grida ] lists the details for the models computed for grid a. grid a consists of 9 models where , in phase i , we vary the following parameters : i ) the type of collapse , ii ) the depletion efficiency , iii ) the final density , and iv ) the initial sulphur elemental abundance . the collapse is either treated as free - fall as described by rawlings et al .
( 1992 ) or retarded .
the depletion efficiency is determined by what fraction of the gas phase material is frozen on to the grains , and undergoes hydrogenation .
several routes of hydrogenation for the most significant species ( o , n , c , co ) have been explored .
the freeze - out fraction is arranged by adjusting the grain surface area per unit volume , and assumes a sticking probability of unity for all species .
the fraction of material on grains is then dependent on the product of the sticking probability and the amount of cross section provided per unit volume by the adopted grain size distribution .
this product was varied so that at the end of phase i , we would have different percentages of ices ( see table [ tb : grida ] ) . note that as the chemistry is time - dependent , different species form at different times and as a consequence the material frozen out on the grains at any one time is @xmath8 representative of the whole gas but of selected species ( in this case
, we chose to monitor frozen co ) .
for the final density , we chose a lower limit of 10@xmath5 @xmath4 and an upper limit of 5@xmath910@xmath7 @xmath4 .
the initial relative abundance of sulphur is very uncertain ( e.g. , ruffle et al .
1999 ) ; we chose the solar value as an upper limit and a factor of hundred lower than the solar value as a lower limit . during phase ii ,
the temperature rises fast and it reaches a maximum of 100 k. .model parameters for grid a. the notation a(b ) signifies a @xmath9 10@xmath10 . the model number is listed in column 1 ; column 2 shows the density of the gas ; column 3 , 4 and 5 shows respectively the percentage of the gas depleted onto grain at the end of phase i , the type of collapse ( b=1 free fall ; b=0.1 retarded ) , the initial sulphur abundance . [
cols="^,^,^,^,^",options="header " , ] [ tb : l1157 ]
we have presented here a detailed time - dependent chemical model of the chemically rich clumps observed along outflows .
this preliminary study was aimed at finding some observable tracers that could help us understanding the origin of the clumps with respect to the outflow .
we also attempted a qualitative comparison of our models with some observations , in particular with the clumps observed along the cb3 outflow ( codella & bachiller 1999 ) . despite large uncertainties in the mode of formation and several chemical assumptions we made
, we believe that we are able to constrain some of physical and chemical parameters of the cb3 clumps .
our conclusions are : 1
. the initial sulphur abundance of the gas forming the clump can not be solar .
we find a depletion factor of @xmath2 100 , confirming the findings of other studies ( oppenheimer & dalgarno , 1974 ; ruffle et al .
2 . a substantial freeze out must occur during the formation of the clump , regardless of its mode of formation .
our models indicate that the most likely explanation for the outflow clumps is that they are pre - existing , meaning @xmath11 that their density structure is , at least partly , formed prior of the advent of the outflow .
this does @xmath8 exclude the general explanation that the outflow clumps are mainly made of swept - up ambient gas and that therefore the clumps are an indication of the episodic nature of the outflows .
it is probable that , with the advent of the outflow , not only the temperature of the clumps increases and reaches the one observed , but the clumps also undergo a period of non - dissociative shock ( and therefore high temperatures , @xmath2 1000 k ) .
the rich chemistry of the clumps observed along cb3 , and l1157 , seems to be a consequence of a pre - existing density enhancement ( either uniform or already in clumps ) and of its interaction with the outflow .
the latter , most likely , shocks and accelerate the gas , and possibly , if episodic , induces its clumpiness .
this is indicated by the high abundance of methanol and some of the sulphur - bearing species .
in fact these molecules are formed by a combination of freeze out and surface reactions , and shocked chemistry : both most efficient when the outflow compresses already dense material at its passage
we find that it is not possible for the outflow clumps to have a uniform high density - a density gradient is needed in order to account for the observed emission of most species . 7 .
co , ch@xmath0oh , cs and so@xmath1 are most likely emitted from the lower density components , while so and ocs come from an intermediate density component .
this chemical stratification supports the findings of bachiller et al .
8 . at late times
( @xmath12 yr ) h@xmath1co is @xmath13 over - abundant by at least a couple of orders of magnitudes .
this discrepancy is similar to that found for the clumps ahead of herbig - haro objects ( viti et al .
2003 ) : these objects however do not share a common chemistry and we find no chemical reason why the abundance of h@xmath1co predicted by our models should be much larger than apparently observed . a possible explanation may be that the observed clumps are smaller than the size implied by the observations .
in fact , when computing the theoretical column densities for smaller sizes ( see section 3.3.4 ) , the match between theory and observations improves . in conclusion , we suggest that interferometric observations of outflow clumps closer to us than cb3 are performed .
this may reveal the real structure of these clumps and help constrain the models .
the authors are indebted to prof d. a. williams for useful conversations about clumpiness and for a critical reading of this revised version .
sv and mb thank the italian space agency for financial support .
sv acknowledges individual financial support from a pparc advanced fellowship .
the authors are very grateful to the referee for comments and suggestions that substantially improved the manuscript .
arce h. g. , & goodman a. a. , 2001 , apj 554 , 132 arce h. g. , & goodman a. a. , 2002 , apj 575 , 928 bachiller r. , 1996 , ara&a 34 , 111 bachiller r. , martn - pintado j. , fuente a. , 1991 , a&a 243 , l21 bachiller r. , p ' erez guti ' errez m. , kumar m. s. n. , tafalla m. , 2001 , a&a 372 , 899 bergin e. a. , neufeld d. a. , melnick g. j. , 1999 , apj 499 , 777 bergin e. a. , alves j. , huard t. , lada c. j. , 2002 , apj 570 , l101 bourke t.l .
, garay g. , lehtinen k.k .
, et al . , 1997 , apj 476 , 781 charnley s. b. , 1997 , apj 481 , 396 codella c , , & bachiller r. , 1999 , a&a 350 , 659 dutrey a. , guilloteau s. , bachiller r. , 1997 , a&a 317 , l55 falle s. a. e. g. , & hartquist t. w. , 2002 , mnras 329 , 195 garrod r. t. , williams d. a. , hartquist t. w. , rawlings j. m. c. , viti s. , 2003 , a&a , submitted hatchell j. , thompson m. a. , millar t. j. , macdonald g. h. , 1998 , a&a 338 , 713 hatchell j. , viti s. , 2002 , a&a 381 , l33 hester j. j. , stapelfeldt k. r. , scowen p. a. , 1998 , aj 116 , 372 le teuff y. h. , millar t. j. , markwick a. j. , 2000 , a&as 146 , 157 millar t. j. , farquhar p. r. a. , willacy k. , 1997 , a&as 121 , 139 millar t. j. , & hatchell j. , 1998 , faraday discussions n. 109,``chemistry and physics of molecules and grains in space '' , p. 15
morata o. , girart j. m. , estalella r. , 2003 , a&a 397 , 181 oppenheimer m. , & dalgarno a. , 1974 , apj 187 , 231 rawlings j. m. c. , hartquist t. w. , menten k. m. , williams d. a. , 1992 , mnras 255 , 471 richer j.s . ,
shepherd d.s .
, cabrit s. , bachiller r. , churchwell e. , 2000 , in protostars and planets iv , ed . mannings v. , boss a.p . ,
russell s.s .
, university of arizona press , p .
867 rowan - robinson m. , 1980 , apjs 44 , 403 ruffle d. p. , hartquist t. w. , caselli p. , williams d. a. , 1999 , mnras 306 , 691 tafalla m. , myers p. c. , caselli p. , walmsley c. m. , comito c. , 2002 , apj 569 , 815 van dishoeck e. w.,1998 , in the molecular astrophysics of stars and galaxies , ed
. t. w. hartquist & d. a. williams ( clarendon press , oxford ) , 53 viti s. , & williams d. a. , 1999 , mnras 310 , 517 viti s. , girart j. m. , garrod r. , williams d. a. , estalella r. , 2003 , a&a 399 , 187 | we investigate the origin of the small , chemically rich molecular clumps observed along the main axis of chemically rich outflows such as cb3 and l1157 . we develop a chemical model where we explore the chemical evolution of these clumps , assuming they are partially pre - existing to the outflow , or alternatively newly formed by the impact of the outflow on the surrounding medium .
the effects of the impact of the outflow are reproduced by density and temperature changes in the clump .
we find that the observed abundances of ch@xmath0oh , so and so@xmath1 are best reproduced by assuming a scenario where the dense molecular gas observed is probably pre - existing in the interstellar medium before the formation of their exciting ( proto)stars and that the clumpiness and the rich chemistry of the clumps are a consequence of a pre - existing density enhancement and of its interaction with the outflow .
[ firstpage ] ism : clouds - ism : jets and outflows - ism : molecules - ism : individual objects : cb3 - ism : individual objects : l1157 |
in a recent article ` coherent lagrangian vortices : the black holes of turbulence ' @xcite the authors draw parallels between certain fluid dynamical configurations and certain properties of black hole geometry .
the goal the article @xcite is to characterize vortex structures which remain coherent over long times .
some such vortices ( known as the agulhas rings ) are observed in the south atlantic and are believed to be relevant for the long range transport of water with relatively high salinity and temperature , and possibly also as moving oases for the food chain ( @xcite and references therein ) .
specifically , the authors of @xcite describe how to associate a 1 + 1 dimensional _
lorentzian _ effective metric tensor to spatial ( constant time ) snapshots of a fluid flow .
one claim made in @xcite is that certain closed spatial curves ( at fixed time ) in the fluid flow are analogous to the ` photon spheres ' that exist around black holes .
a photon sphere necessarily occurs in a black hole space - time and here we draw intuition ( for a subsequent two dimensional treatment ) from the simplest black hole , which is described ( in four dimensions ) by the schwarzschild metric ( see below ) . while we we do not take issue with the interesting fluid - dynamical subject matter of @xcite , we wish to point out a number of conceptual difficulties associated with the geometric interpretation of the results and hopefully elucidate some of the ( perhaps counter - intuitive ) features of lorentzian geometry and in particular of black hole geometries .
specifically , we make the following clarifications : * the circular photon orbit ( associated with the photon sphere ) around a schwarzschild black hole is _ not _ a closed null geodesic .
* closed null curves in general relativity are extremely pathological and probably forbidden by reasonable physics arguments ( for example , globally hyperbolic spacetimes do not admit closed null curves ) . *
the existence of a photon sphere is not a necessary nor is it a sufficient condition for the existence of a black hole . *
the circular photon orbit is _ not _ circular - its 3-dimensional _ projection onto the fixed time slices of the schwarzschild geometry in spherical polar coordinates _ is . * a singularity in the metric
does not imply a singularity of the geometry .
a singularity of a metric coefficient in one coordinate system is a weak condition and does not indicate the presence of a real physical singularity . below we will expand on these points . in sec .
[ s:2 ] we explore the construction of the ` fluid - metric ' as defined in @xcite and its geometry . in sec .
[ s:3 ] we review some relevant aspects of the schwarzschild geometry , in particular the circular photon orbit and the photon sphere , while in sec . [ s:4 ] we present some examples of fluid flows which give rise to interesting lorentzian geometries and which serve to illustrate our observations .
in this section we briefly review the main theoretical result of ref .
@xcite by deriving the @xmath0-dimensional lorentzian metric associated with certain fluid configurations .
the primary mathematical result of the article @xcite is the identification and characterisation of certain closed curves along which the fluid flow is ` coherent ' : closed curves of material flow which remain closely associated .
such curves are shown in @xcite to satisfy a differential equation which is interpreted as an integrability condition of a vector field which is null with respect to an auxiliary metric tensor .
that is , one discovers a null vector field and subsequently integrates it to yield the integral curves which are null curves .
we shall introduce some concepts from continuum mechanics to this end .
consider a sufficiently smooth fluid flow velocity @xmath1 ( for example , @xmath2 could be a solution of the navier - stokes equations , but this is not required ) .
then a fluid element propagates in time along the flow lines with a world line @xmath3 which satisfies = ( t , ) .
[ e : ode ] the flow given by represents a map @xmath4 from @xmath5 to @xmath5 ( we restrict attention to a 2-dimensional fluid flow ) where the reference configuration given by the initial condition @xmath6 gets mapped to the solution to at time @xmath7 _ t:_0 ( t ) .
here we have denoted the coordinates in the plane at time @xmath7 with a capital letter @xmath8 for clarity while reserving the lower case variable names for the initial un - deformed coordinates @xmath9 . in general , the flow @xmath4 will deform the lines of constant initial coordinate label , say @xmath10const . or @xmath11const . , where @xmath12 are cartesian coordinates , so that lines of constant initial condition @xmath13 or @xmath14 will , at some fixed time @xmath7 , be deformed to some curved lines in the plane .
we may characterise such a deformation with the so - called ` deformation gradient ' mixed two - point tensor @xmath15 whose components with respect to an initial and final coordinate basis are given by f^i _ j:= .
note the mixed - coordinate definition ( two - point tensor structure ) by observing that these components are given with respect to the two coordinate bases , initial and final in a specific way .
we have , in coordinate - independent notation , = f^i_j ( dx^j ) . for example , and for later reference , in a polar coordinate basis we have f^i_j= & + + & , which is written in the more convenient orthonormal polar basis @xmath16 and @xmath17 ( @xmath18 and @xmath19 ) as f^a_b= & + + r(t ) & .
the orthonormal basis is preferable to the polar coordinate basis due to the possibility to raise and lower indices with the identity matrix , which coincides with the matrix of metric coefficients in such a basis . the `
right cauchy - green ' deformation tensor @xmath20 carries the same deformation information modulo transformations which merely shift or rotate an initial configuration without genuine deformation and is defined as c^i_j:=f_k^i f^k_j = g_km g^in , where @xmath21 and @xmath22 are the metric coefficients in the two coordinate bases , respectively . in orthonormal bases
we have c_ab=_c .
we will be concerned with the eigenvectors @xmath23 and the eigenvalues @xmath24 of @xmath20 .
note that @xmath20 as well as the eigenvectors and eigenvalues depend on the initial point @xmath25 as well as on the time @xmath7 at which we calculate the strains .
being the ` square ' of another tensor @xmath15 it is simple to show that the eigenvalues @xmath26 of @xmath20 satisfy 0_1(_0,t)_2(_0,t ) _ 0 , t and that @xmath27 form an orthonormal basis for @xmath5 .
let @xmath28 and define the so - called ` generalised green - lagrange tensor ' @xmath29 in an orthonormal frame by e_ab:=. [ e : e ] this object can be thought of as acting as a time - dependent bilinear form on the initial conditions space and depends on the constant @xmath30 . in the ` moving orthonormal frame ' defined by the eigenvectors of @xmath20 we can write e_ab= _ 1- & 0 + 0 & _ 2- .
[ e : eeigen ] the authors of @xcite define an effective metric tensor induced by the action of @xmath31 in on the space of initial conditions which , remarkably , can have lorentzian signature ( although the tensor @xmath31 is of the wrong type to be a metric , we gloss over this and talk about @xmath31 as ` being ' the metric tensor ) . in principle
, this fact allows one to draw parallels between this structure and a black hole metric with an associated photon sphere . the first thing to notice about the tensor
@xmath31 is that its signature can change from point to point and over time depending on the relative magnitude of the eigenvalues @xmath26 and is only lorentzian at time @xmath7 if @xmath33 for all points @xmath25 . assuming this ` lorentzian condition ' on the eigenvalues at some time @xmath7
, we can construct the two independent ` e - null ' vector fields @xmath34 in the orthonormal eigen - basis of @xmath20 as @xcite @xmath35 where we have ( arbitrarily but without loss of generality ) normalised the vector fields to have unit norm in the background euclidean metric .
the integral curves of these vector fields are null curves with respect to the metric @xmath31 .
it is shown in @xcite that these null curves are curves for which the flow up to time @xmath7 _ uniformly _ scales the tangent vectors to the curve .
that is , given an @xmath31-null curve @xmath36 which gets mapped under the flow @xmath4 to the curve @xmath37 , we have ||_t()(s)||^2=||(s)||^2 uniformly for all @xmath38 parametrizing the curve and where the prime denotes taking the tangent vector at the given point on the curve . such closed curves which possess the uniform stretching property are intuitively ` invariant curves ' of the flow @xmath4 ; their fluid dynamical relevance is discussed in @xcite where the outermost curve in a family of such curves is said to define the boundary of a ` coherent material vorticex ' in 2-dimensional flow .
the authors of @xcite refer to a photon sphere in an analogy with vortices through the lorentzian metric @xmath31 .
a photon sphere is associated with a black hole spacetime metric . in general relativity
there exists a vast array of black hole and black hole - like solutions with varying degrees of physical sensibility . in the vacuum ,
spherically symmetric case in @xmath39-dimensions , there is a uniqueness theorem ( birkoff s theorem @xcite ) which points us to the static schwarzschild black hole metric [ schwarzschild ] ds^2=-(1-)dt^2 + ( 1-)^-1dr^2+r^2 d_(2)^2 here written in standard spherical polar coordinates , where @xmath40 is the metric on the unit 2-sphere , @xmath41 is newton s constant , @xmath42 is the black hole mass , and units in which the speed of light is unity are used , as is common in relativity .
the metric ( [ schwarzschild ] ) describes the simplest black hole spacetime in @xmath39 dimensions .
a crucial feature of this metric is the existence of an horizon a lower - dimensional region on which the metric is degenerate and at which time and space ` swap their roles ' . without getting into technical details ,
a precise statement is that the norm of the timelike killing vector field associated with the time symmetry changes sign , becoming a space - like vector field at a co - dimension 2 hyper - surface generated by null geodesics of the metric ( [ schwarzschild ] ) and known as the _ event horizon_. the event horizon is the key structure which , if present in a gravitational configuration , indicates the presence of a black hole . in the above schwarzschild case ,
the time - like killing vector is simply the time direction @xmath43 , with norm ||k_t||^2=-(1- ) which changes sign at @xmath44 , a value of the radial coordinate known as the _ schwarzschild radius _ or black hole radius @xcite .
this is the radius of the event horizon . without an event horizon
, a spacetime can not be sensibly said to contain a black hole . in a careful relativistic treatment @xcite
, one can show that the effective potential for ` free ' ( geodesic ) radial motion in the schwarzschild geometry is given by the effective 1-dimensional potential [ effectivepotential ] v(r)=- + - , where @xmath45 for massive particles and @xmath46 for massless particles , and @xmath47 is the angular momentum represents the angular momentum per unit mass .
] of the particle .
the first two terms on the right hand side of eq .
( [ effectivepotential ] ) are identical to the two terms which appear in the effective potential of the newtonian treatment @xcite , while the third one is a correction due to general relativity .
circular orbits exist at radii @xmath48 for which @xmath49 and we have r_=(l ) . in the massless limit @xmath46
, there exists a single orbit at @xmath50 , which is unstable , while in the massive case there exist two orbits , one stable and the other ( at smaller radius ) unstable .
in contrast , in newtonian gravity ( in which the term proportional to @xmath51 is absent in the effective potential @xmath52 ) , there exists a single stable circular orbit for massive particles and no circular orbit for massless particles .
it is important to note that this closed circular orbit is _ not _ a closed null geodesic , such curves being highly pathological and most likely unphysical in general relativity .
indeed , it is not even a geodesic nor is it a null curve .
instead it is the spatial projection of an open and infinitely extended null geodesic curve in space and time as shown in fig .
[ f : helix ] . ] the tangent vector to the null geodesic is given by = ( 3e,0,0 , ) in the spherical coordinate basis whose integral curve is the helical null geodesic in spacetime x^()=(3e , 0,0 ) .
a photon sphere is a sphere of radius equal to that of the closed circular photon orbit and coincides with a 2-sphere of symmetry of the spacetime .
the photon sphere ( a spacelike hypersurface ) of radius @xmath50 has nothing to do with the black hole horizon ( a null hypersurface ) of radius @xmath53 , which traps all particles .
in this section we consider various background fluid flows and compute the associated null curves of the auxiliary metric as defined in .
the purpose of this section is to use examples to highlight the distinction between the closed null curves and any kind of photon sphere . in an attempt to construct simple examples possessing closed curves uniformly stretched by a flow
, we consider rotational symmetry .
it is clear that circles concentric about the origin are uniformly stretched in circularly symmetric flows and hence constitute examples of the curves sought in ref .
@xcite as solutions to the optimization problem and shown to define ` coherent material vortices ' .
we consider sequentially more complex flows : rigid body , irrotational , and some rotating and draining vortex flow .
consider a rigid body flow profile _ t:(r_0,_0)(r_0 , _
0+t ) where @xmath54 is a fixed angular frequency .
then @xmath55 coincides with the identity , @xmath56 , and the effective metric @xmath29 is not of lorentzian signature for any @xmath30 .
hence we must move to a more complicated flow in order to explore the construction . in a certain sense , is null in this geometry , highlighting the total uniform and coherent ( non - deforming ) nature of the flow .
such a flow is highly unphysical and does in no way invalidate the results of @xcite : it serves only as a first simple example which shows that the metric construction can be non - trivial .
consider instead the differential rotating flow _ t:(r_0,_0)(r_0 , _ 0 + ) .
[ e : irrot ] such a flow is irrotational and is commonly refered to as a ` vortex line ' flow , in this case of strength @xmath57 .
then the eigenvalues of @xmath20 are easily calculable as _ = and we see that they are independent of the angular variable @xmath58 .
they satisfy , as functions of @xmath7 and @xmath59 , the bounds @xmath60 as well as the bounds at fixed @xmath7 @xmath61 ( see fig .
[ f : values ] ) .
hence the only value of @xmath30 for which @xmath29 is of lorentzian signature in the entire plane is @xmath62 . of the matrix @xmath20 as a function of @xmath59 at finite time @xmath7 for the irrotational vortex flow .
as time proceeds the red solid curve diverges to @xmath63 while the blue dashed curve converges to @xmath64 neither curve crosses the black dot - dashed line @xmath62 at any time @xmath7 .
[ f : values ] ] associated with the irrotational vortex flow .
this vector field is independentr of the time @xmath7 up to which the flow is computed .
[ f : nulls1 ] ] [ e : eta- ] ( here at time @xmath65 ) associated with the irrotational flow .
this additional null vector field depends on the time @xmath7 up to which we compute the flow . at late times , or small radius
it converges to @xmath66 , i.e. tangent to concentric circles about the origin while at early times it is purely radially pointing .
[ f : nulls2 ] ] in the orthonormal polar basis for this choice of @xmath30 we have e_ab= ( ) ^2 & - + - & 0 in this case , one of the two null vector fields of @xmath31 is time - independent and tangent to circles concentric on the origin ( see fig . [
f : nulls1 ] ) , in the orthonormal polar basis , _ + = ( 0,1 ) while the other is time - dependent , _
-= ( 1 , ) [ e : eta- ] tending to a purely angular pointing ( in the @xmath67 direction ) vector field as @xmath68 ( see fig .
[ f : nulls2 ] ) .
it is straightforward to show that the geodesic equation ( one might find it easier to work in the non - orthonormal polar coordinate basis at this stage ) becomes the pair @xmath69 from which we see that our null curves of constant @xmath70 are indeed null geodesics .
furthermore one can simply show that also the second set of null curves are geodesics . using standard techniques
one can show that there are three killing vector fields , two of which are ` spacelike ' for all @xmath7 and @xmath25 and one of which is null for all @xmath7 and @xmath25 . here ` spacelike ' is an arbitrary definition which we take to mean positive norm ( with respect to the metric @xmath31 ) .
the null killing vector field is purely rotational being tangent to circles about the origin . since there is no 1-dimensional line on which any of the killing vectors change norm we can conclude that there does not exist a horizon in this geometry .
again , this flow is highly idealised being unbounded at the origin whereas in practice when working with real turbulent flows one would have a region @xmath71 which might even be time dependent .
one might consider also a _ draining _ vortex type fluid flow _ t:(r_0,_0 ) ( ( t)r_0 , _ 0 + ) .
where @xmath72 is a function which parametrises the radial flow , in the hopes of unearthing some lorentzian geometry which more closely resembles that of a black hole .
indeed , with the addition of radial flow one might hope to have some kind of `` trapped region '' inside of which all time - like vectors point towards the origin , reminiscent of a similar construction in the singularity theorems of hawking and penrose@xcite .
for example one might choose @xmath72 such that the flow describes a draining ` bathtub vortex ' which to a first order approximation can be described by @xmath73 so that @xmath74 and the radial velocity is inversely proportional to the radial position as a function of time .
the flow and geometry given here are more complex than in the previous example and , for brevity , we shall not present them in any depth here .
it can be shown , however , that at fixed time @xmath7 the matrix @xmath20 in this case has two eigenvalues which behave analogously to the irrotational vortex case but separated now by the time - dependent ` constant ' @xmath75 . in this case , and in line with our intuition , circles concentric about the origin are invariant curves and are null geodesics of the metric @xmath76 : a circle of radius @xmath59 gets mapped under the flow to circles of radius @xmath77 so that the tangent vectors squared are uniformly scaled by @xmath78 .
this is intuitive since an experimentalist who drops ink droplets in a perfect ring into such a flow will observe a shrinking of the ring ( or an expansion , depending on the character of @xmath79 ) but it will be _ coherent _ ( not deforming ) as time progresses .
while the characterisation of fluid vortices in terms of an auxiliary lorentzian metric on fixed time slices of a fluid flow should be of practical utility , as pointed out in @xcite , the interpretation of the metric and its lorentzian geometry as being ` close to ' or analogous to that of a black hole is lacking .
we have shown that no event horizons exist for simple flows which possess the characteristic features discussed in @xcite as being ` photon sphere ' or ` black - hole - like ' .
further , we have clarified the nature of the circular photon orbit in the schwarzschild geometry ( the prototypical black hole geometry containing an horizon surface which traps massive and massless particles ) ) and shown it to be very different from the closed null curves which hearald the ` coherent material vortices ' discussed in @xcite .
specifically the @xmath30-lines discussed in @xcite as being analogous to the photon sphere are closed null curves while the photon sphere contains closed space - like curves .
it is interesting to point out that the use of lorentzian geometry and auxiliary metrics in fluid dynamics in fact has a healthy and vibrant research community and literature associated with it known as the ` analogue gravity ' program @xcite where the effective metric discussed there has a sound physical basis , being the metric which describes the real causal structure of a physical system for various kinds of physical propagating signals such as sound waves or small perturbations . in this manner , the metrics in analogue gravity are expected to be sufficiently ` physical ' , for example they can not contain closed null or closed time - like curves .
the cutting edge of that discipline is the construction and observation of analogue event horizons or analogue photon spheres in analogue systems ( see for example @xcite ) . in still other corners of the literature
the introduction of an effective metric in a fictitious space is the key feature of the jacobi form of the maupertuis variational principle in point particle mechanics @xcite . while in principle intriguing , the introduction of such effective metrics in that context ( see also @xcite for effective metrics in different contexts ) has , thus far , been of little utility in developing theories and solving practical problems .
it is hoped that the characterization of eddies by means of an effective lorentzian geometry will break this spell .
we thank g. haller and f.j .
beron - vera for comments on a previous version of the manuscript .
this work is supported by the natural sciences and engineering research council of canada and by bishop s university .
g. haller and f. j. beron - vera , `` coherent lagrangian vortices : the black holes of turbulence , '' _ j. fluid mech .
_ * 731 * ( 2013 ) r4 , http://arxiv.org/abs/1308.2352[arxiv:1308.2352 [ physics.ao-ph ] ] .
s. m. carroll , _ spacetime and geometry_. addison - wesley , san francisco , 2005 .
s. hawking and r. penrose , `` the singularities of gravitational collapse and cosmology , '' _ proc.roy.soc.lond . _ * a314 * ( 1970 ) 529548 . c. barcelo , s. liberati , and m. visser , `` analogue gravity , '' _ living reviews in relativity _
* 14 * ( 2011 ) , no . 3 , .
g. rousseaux , p. maissa , c. mathis , p. coullet , t. g. philbin , _ et al .
_ , `` horizon effects with surface waves on moving water , '' _ new j.phys._ * 12 * ( 2010 ) 095018 , http://arxiv.org/abs/1004.5546[arxiv:1004.5546 [ gr - qc ] ] .
s. weinfurtner , e. w. tedford , m. c. penrice , w. g. unruh , and g. a. lawrence , `` classical aspects of hawking radiation verified in analogue gravity experiment , '' _ lect.notes phys . _
* 870 * ( 2013 ) 167180 .
t. g. philbin , c. kuklewicz , s. robertson , s. hill , f. konig , _ et al .
_ , `` fiber - optical analogue of the event horizon , '' _ science _ * 319 * ( 2008 ) 13671370 , http://arxiv.org/abs/0711.4796[arxiv:0711.4796 [ gr - qc ] ] .
a. n. kolmogorov in _ proceedings of the international congress on mathematics _ ,
north holland , amsterdam , 1954 .
n. abe , `` notes on the kolmogorov s remark concerning classical dynamical systems on closed surfaces , '' in _ geometry of geodesics and related topics , advanced studies in pure mathematics 3 _ , k. shiohama , ed . north holland , amsterdam , 1984 . | in this letter we point out some interpretational difficulties associated with concepts from general relativity in a recent article which appeared in _ j. fluid mech . _ ( 2013 ) r4 where a lorentzian metric was defined for turbulent fluid flow and interpreted as being analogous to a black hole metric .
we show that the similarity with black hole geometry is superficial at best while clarifying the nature of the black hole geometry and the work above with some examples . |
in @xcite , h. gangl , d. zagier and the first author studied in detail the `` double shuffle relations '' satisfied by the double zeta values @xmath0 and revealed in particular various connections between the space of double zeta values and the space of modular forms as well as their period polynomials on the full modular group @xmath1 .
they also defined the `` double eisenstein series '' and deduced the double shuffle relations for them , and in @xcite we illustrated a direct way to connect the double eisenstein series to the period polynomials of modular forms ( of level 1 ) . in the present paper , we consider the double shuffle relations of level @xmath2 and study the formal double zeta space , whose generators are the formal symbols corresponding to the double zeta values of level 2 ( euler sums ) and the defining relations are the double shuffle relations .
one of the relations we obtain in the formal double zeta space ( theorem 1 ) has an interesting application to the problem of representations of integers as sums of squares , and this will be given in the subsequent paper by the second author @xcite .
we then proceed to define the double eisenstein series of level 2 and show that they also satisfy the double shuffle relations ( theorem 3 ) , and have connections like in the case of level 1 to double zeta values , modular forms , and period polynomials , of level 2 ( theorem 5 and its corollary ) .
this work is partially supported by japan society for the promotion of science , grant - in - aid for scientific research ( s ) 19104002 , ( b ) 23340010d
the double zeta values of level @xmath2 we are referring to are the following four types of real numbers given for integers @xmath3 and @xmath4 : @xmath5 these numbers can be written as simple linear combinations of the original multiple zeta values and the numbers often referred to as euler sums defined by @xmath6 and vice versa .
we have for instance @xmath7 and similarly for other values .
note that , from the obvious relation @xmath8 and @xmath9 we have the relation @xmath10 we shall hereafter only consider @xmath11 , and @xmath12 moreover define @xmath13 then in the standard manner we can show the following double shuffle relations . for positive integers @xmath14 , we have @xmath15 the first equality in each sequence of identities is obtained as usual from the manipulation of the defining series . for the second , we use the following integral representations of each zeta value and the shuffle product of integrals : @xmath16 the first two are easy to deduce , and to see the rest for double zetas , we use the expressions of those values in terms of the multiple @xmath17-values studied in @xcite : @xmath18 here , the multiple @xmath17-value ( of level 2 ) @xmath19 is so defined that it has an integral expression similar to that for usual multiple zeta values , and in our case ( @xmath20=2 ) it is given as @xmath21 where @xmath22 or @xmath23 .
the value @xmath24 is nothing but the double zeta value @xmath25 . in terms of the euler sums
mentioned before , we have @xmath26 noting the identities @xmath27 we obtain the desired integral expressions and hence the proposition by shuffle products .
now we introduce the level 2 version of the formal double zeta space studied in @xcite as follows .
let @xmath28 and @xmath29 be the @xmath30-vector space spanned by formal symbols @xmath31 , @xmath32 ,
@xmath33 , @xmath34 , and @xmath35 with the set of relations @xmath36 for @xmath37 , so that @xmath38 since the elements @xmath33 and @xmath39 are written in @xmath40 s , we can also regard the space as given by @xmath41 where the defining relations and are @xmath42 note that the relations and ( as well as and ) correspond to those in proposition 1 when @xmath43 , under the correspondences @xmath44 because in that case the binomial coefficients for @xmath45 on the right vanishes . for our later applications
it is convenient to allow the `` divergent '' @xmath46 etc . , and
in fact the double shuffle relations in proposition 1 can be extended for @xmath47 or @xmath48 by using a suitable regularization procedure for @xmath49 etc . developed in @xcite .
specifically , by setting @xmath50 @xmath51 and , for @xmath52 @xmath53 where @xmath54 is a formal variable , the equations in proposition 1 are valid for all @xmath55 except @xmath56 .
suppose @xmath57 is even and @xmath58 . in @xmath29
, we have \1 ) @xmath59 \2 ) each @xmath60 with @xmath61 even can be written as a @xmath30-linear combination of @xmath62 and @xmath35 consider the generating functions @xmath63 here and in the following , the sum @xmath64 always means @xmath65
. the double shuffle relations and are equivalent to the relations @xmath66 substituting @xmath67 in and @xmath68 in , we respectively obtain @xmath69 we divide by 2 and add to obtain @xmath70 and hence 1 ) of theorem . to prove 2 ) ,
we need the following lemma . [
10 ] let @xmath71 be an even integer and @xmath72 be rational numbers
. then the following two statements are equivalent .
\1 ) the relation @xmath73 holds in @xmath29 ( as before @xmath74 means @xmath75 ) .
\2 ) there exist some homogeneous polynomials @xmath76 $ ] of degree @xmath77 such that @xmath78 this is an analogue of proposition 5.1 in @xcite .
take @xmath79 ( and @xmath80 ) and compute the coefficients of @xmath81 using binomial theorem .
then the relation in 1 ) is exactly ( not only @xmath82 but as an exact equality ) the relation .
similarly , by taking @xmath83 ( and @xmath84 ) and computing the coefficients of @xmath85 , we see that the relation in 1 ) is the relation modulo @xmath86 .
since any relation of the form in 1 ) in @xmath29 should come from a linear combination of and modulo @xmath86 , and any homogeneous polynomial is a linear combination of monomials , we obtain the lemma . using the lemma
, we are going to produce enough relations of the form @xmath87 such that we can solve these in @xmath33 . in view of the relations
@xmath88 and the lemma , we obtain the relation of the form if we can take @xmath89 and @xmath90 in 2 ) of lemma 1 so that the coefficients satisfy 1 .
@xmath91 , 2 .
@xmath92 , 3 .
@xmath93 for all odd @xmath94 .
we now work for convenience with inhomogeneous polynomials .
recall the usual correspondences @xmath95 and @xmath96 , and the action of the group @xmath97 on the space of polynomials of degree at most @xmath77 by
@xmath98 we extend this action to the group ring @xmath99 $ ] by linearity .
set @xmath100 then the left - hand side of the equation in 2 ) of lemma 1 can be written in inhomogeneous form as @xmath101 ( we write @xmath102 instead of @xmath103 . )
suppose the polynomial @xmath104 ( of degree at most @xmath77 ) satisfies @xmath105 and put @xmath106 .
then the expression gives the coefficients ( in lemma 1 - 2 ) ) satisfying the above three conditions ( i ) , ( ii ) , ( iii ) .
inserting @xmath106 into and using the assumption @xmath105 , which is equivalent to @xmath107 since @xmath108 , and also using the identities @xmath109 in @xmath110 , we can write as @xmath111 now the condition ( iii ) ( the polynomial is even ) is clear from this ( being killed by @xmath112 ) , and the conditions ( i ) and ( ii ) are respectively the consequences of the equations @xmath113 noting @xmath114 and hence @xmath115 we see that the polynomials @xmath116 for @xmath117 ( even ) satisfy the condition @xmath118 in lemma 2 . with this choice of @xmath119 ( for @xmath120 ) and @xmath121 in lemma 2 , we compute the coefficients in lemma 1 by noting , and by using @xmath122 to obtain a relation of the form @xmath123 when we put @xmath124 , we can solve these congruences successively in each @xmath125 for @xmath126 .
this completes the proof of theorem 1 .
we introduce the double eisenstein series of level 2 and first show that they satisfy the double shuffle relations .
let @xmath127 ( resp .
@xmath128 ) be the set of even ( resp .
odd ) integers and @xmath129 a variable in the upper half - plane .
define the three double eisenstein series @xmath130 , and @xmath131 by @xmath132 here , the positivity @xmath133 of a lattice point means either @xmath134 or @xmath135 , and @xmath136 means @xmath137 .
we assume @xmath138 and @xmath52 for the absolute convergence .
all the series in is easily seen to be invariant the translation @xmath139 , and hence have fourier expansions .
the fourier series developments can be deduced in a quite similar manner to the full modular case @xcite . [
6 ] let @xmath140 and @xmath141 be integers and set @xmath142 .
we have the following @xmath143-series expansions ( @xmath144 ) .
@xmath145 where @xmath146 is kronecker s delta , @xmath147 and @xmath148 @xmath149 , and the @xmath121 s are the following @xmath143-series : @xmath150 and @xmath151 put , for positive integers @xmath61 and @xmath152 , @xmath153 then the series @xmath154 in the theorem can be written by using @xmath155 as @xmath156 the computation of the fourier series can be carried out in a completely similar fashion as done in @xcite , dividing the sum of the defining series into four terms , according as @xmath157 .
for instance , in the case of @xmath158 , we compute @xmath159 using the partial fraction decomposition @xmath160 and the formulas @xmath161 ( consequences of the standard lipschitz formula , and when @xmath47 we use @xmath162 instead ) .
we leave the details to the reader .
we remark that each series in theorem 2 is in @xmath163+\sqrt{-1}{\mathbb r}[[q]]$ ] , and the terms in @xmath164 $ ] ( `` imaginary part '' ) only come from the terms having @xmath165 with odd @xmath166 as coefficients .
now we extend the definition of the double eisenstein series for any ( non - converging ) @xmath167 ( except @xmath168 ) , by using @xmath143-series . for this
, we separately define the imaginary part and the `` combinatorial part '' .
first we define the imaginary parts as @xmath169 the sum is over @xmath170 with @xmath166 odd .
note that the regularized values @xmath171 and @xmath172 are defined by and thus for any positive integers @xmath173 , these series are elements of @xmath174[[q]]$ ] .
secondly , we define the part in @xmath175 $ ] which is referred to as the combinatorial double eisenstein series . put @xmath176 where @xmath177 and as before the condition `` @xmath178 '' includes `` @xmath179 '' .
let @xmath180 and for integers @xmath181 let @xmath182 where @xmath183 note that each @xmath184 is @xmath185 when @xmath186 and @xmath141 .
the combinatorial double eisenstein series are then defined , for positive integers @xmath181 , by @xmath187
lastly , the constant term of the double eisenstein series is given by the ( regularized ) double zeta values .
[ 15 ] for any integers @xmath181 with @xmath188 , we define @xmath189 to state the double shuffle relations in the forms and for these series , we need usual eisenstein series for the congruence subgroup @xmath190 . for each integer @xmath191 ,
let the series @xmath192 and @xmath193 be defined by @xmath194 and @xmath195 when @xmath71 is even , the functions @xmath192 and @xmath193 are the eisenstein series for @xmath196 associated to cusps @xmath197 and @xmath185 respectively , and as such they are modular of weight @xmath57 with respect to @xmath196 . the fourier series of @xmath192 and @xmath193 are given as follows .
let @xmath198 be the eisenstein series of weight @xmath57 for @xmath199 : @xmath200 ( note that this gives a non - zero function even when @xmath57 is odd . ) with this we have @xmath201 we define the @xmath143-series @xmath202 , and @xmath203 for any @xmath204 by the ( convergent ) @xmath143-series on the right - hand sides of and , with the regularized values and @xmath205 .
finally we set @xmath206 [ 3_3 ] for any integers @xmath181 with @xmath207 , we have @xmath208 the proof of the theorem will be postponed to [ proof ] . in this subsection , we describe a mysterious connection between our double eisenstein series and the period polynomials associated to cusp forms on @xmath196 .
this kind of connection was first observed in the full modular case @xcite , which will be recalled briefly in the appendix [ appendix ] for the convenience of the reader because the reference @xcite circulated only among participants of the conference .
let us recall the theory of period polynomials for @xmath196 given in @xcite and @xcite .
we follow the formulation of @xcite .
recall the group @xmath196 is generated by two elements ( see e.g. ( * ? ? ?
* theorem 4.3 ) ) ) @xmath209 let @xmath57 be a positive even integer and @xmath210 be the space of polynomials with rational coefficients of degree at most @xmath77 : @xmath211\,\vert\ , \deg ( f)\le k-2\right\}.\ ] ] the group @xmath196 acts on @xmath210 as in ( [ action ] ) and this action extends to that of the group ring @xmath212 $ ] as usual .
consider the subspace @xmath213 of @xmath210 defined by @xmath214 for a cusp form @xmath215 , the period polynomial @xmath216 is given by @xmath217 it is implicitly shown in the proof of proposition 3 in @xcite that @xmath218 now we consider the even and odd parts of polynomials separately .
put @xmath219 . by the identity @xmath220 ( every matrix identity
is regarded projectively , i.e. , as that in @xmath221 ) , we see that if @xmath222 then @xmath223 , and so we have the direct sum decomposition @xmath224 where @xmath225 ( resp .
@xmath226 ) is the even ( resp .
odd ) part of @xmath213 : @xmath227 we also denote by @xmath228 the even and odd part of @xmath216 , @xmath229 and by @xmath230 the map @xmath231 for the space @xmath225 of even polynomials , we have two obvious elements @xmath23 and @xmath232 .
this is clear for @xmath23 because @xmath233 . for @xmath232
, we note the identity @xmath234 and @xmath235 and thus @xmath236 .
hence , we have the decomposition @xmath237 where @xmath238 let @xmath239 be the map @xmath240 obtained by the composition of @xmath241 and the natural projection @xmath242 . from the works of imamoglu - kohnen @xcite and fukuhara - yang @xcite , we obtain the following for even @xmath57 , the two maps @xmath243 are isomorphisms of vector spaces . we know from @xcite and @xcite that both maps are injective .
so all we have to show is the dimensions of the target spaces are equal to the dimension of @xmath244 , which is equal to @xmath245 - 1 $ ] .
we only calculate the dimension of @xmath246 , since the other is similar and only the former is relevant to the subsequent story involving the double eisenstein series
. put @xmath247 .
obviously @xmath248 is equivalent to @xmath249 . for an even polynomial @xmath250 with no constant term and no @xmath232 term
, we compute the condition @xmath251 for @xmath252 being in @xmath246 . by @xmath253 and
@xmath254 the condition becomes @xmath255 which is written as @xmath256 using binomial theorem , we can write this as @xmath257 therefore , the space @xmath246 is the set of polynomials @xmath250 whose ( rational ) coefficients satisfy a set of linear relations @xmath258 clearly the equations for @xmath259 and @xmath260 have just opposite sign , and so we have to look only at the equations for @xmath261 .
but then for @xmath262 the coefficient of @xmath263 is zero , and the coefficient matrix is upper triangular with non - zero diagonals .
we thus see that the rank of this matrix is @xmath264 - 1 $ ] , and the dimension of @xmath246 is @xmath265 - 1\right)=[k/4]-1 $ ] , as desired
. * remark .
* we have not succeeded to characterize the ( codimension 2 ) image of @xmath244 by @xmath241 in @xmath266 .
it is expected that such a characterization as in @xcite should exist .
+ amasingly enough , the coefficient matrix in appears exactly when we look at the imaginary part of the double eisensten series of level 2 , which we are going to explain .
we look at the imaginary part of @xmath267 ( as @xmath143-series ) for @xmath61 even .
let @xmath268 \longrightarrow \sqrt{-1}{\mathbb r}[[q]]$ ] be the natural projection to imaginary part ( note that by imaginary part we mean the term in @xmath164 $ ] , not the coefficient of @xmath269 ) .
as theorem 2 and definition 1 ( for @xmath270 ) shows , imaginary parts come from the terms with @xmath271 for odd @xmath166 , and we have in matrix form @xmath272 where @xmath273 is the @xmath274 matrix given by @xmath275 this is exactly the coefficient matrix of !
let @xmath276 be the @xmath30-vector space generated by @xmath277 .
let @xmath58 be a positive even integer .
\1 ) @xmath278 so that the series @xmath267 @xmath279 even@xmath280 are linearly independent over @xmath30 .
\2 ) the space @xmath276 contains @xmath281 , where @xmath282 is the space of cusp forms on @xmath196 having rational fourier coefficients .
we first prove 2 ) .
we know from theorem 3 and theorem 1 that the space @xmath276 contains @xmath283 , as well as @xmath284 and @xmath285 ( @xmath286 ) .
because of the relation @xmath287 and the fact shown by imamoglu and kohnen in @xcite that these cusp forms @xmath288 generate the space @xmath244 , we obtain the assertion 2 ) . for 1 )
, first we note by definition the inequality @xmath289 since elements in @xmath290 has no imaginary parts , they sit in the kernel of the projection @xmath291 from @xmath276 to @xmath164 $ ] , thus @xmath292 .\ ] ] as for the dimension of the image of @xmath291 , we see that it is equal to the rank of the matrix @xmath273 because the series @xmath293 are linearly independent over @xmath294 .
this can be seen as follows . for a prime @xmath166 , the coefficient of @xmath295 in @xmath296 is essentially @xmath297 .
hence by picking distinct prime numbers @xmath298 and looking at the coefficients of @xmath299 in @xmath293 , we see the desired linear independence because the coefficient matrix is essentially the vandermond determinant .
we thus have @xmath300 -1\ ] ] and therefore @xmath301+\left[\frac{k+2}{4}\right ] -1=\frac{k}{2}-1.\ ] ] therefore we conclude @xmath302 and also @xmath303 [ 5 ] for an even integer @xmath28 , we have @xmath304 by taking the constant term of the @xmath143-series , we obtain the surjective map @xmath305 by the theorem , the kernel of @xmath306 contains the space @xmath282 and hence we obtain the corollary . * remark .
* the above corollary says that among the @xmath307 numbers @xmath308 there are at least @xmath309 linear relations .
it seems that the @xmath307 numbers @xmath310 are linearly independent over @xmath30 , and the total space @xmath311 is spanned by @xmath310 and @xmath312 .
( recall the sum formula in theorem 1 , so that @xmath312 is contained in the above space . )
the conjectural dimension of this space is thus @xmath313 .
we also conjecture that the space of usual double zeta values of even weight @xmath57 is containend in the space spanned by @xmath314 except @xmath315 : @xmath316 and that the space @xmath317 is contained in the usual double zeta space : @xmath318 when @xmath57 is odd , we can prove that every @xmath314 except @xmath315 is a linear combination of @xmath319 : @xmath320 to prove this we use the identity of y. komori , k. matsumoto , and h. tsumura @xmath321 valid when @xmath43 and @xmath322 odd , where @xmath323 for @xmath315 , it seems we need @xmath324 other than usual double zeta values , but we have not proved this . as in @xcite , we prove theorem 3 by dividing it into three parts : the constant term , the imaginary part , and the combinatorial part . the double shuffle relation of the constant term is nothing but that of double zeta values , namely proposition 1 and its regularization . as for the imaginary part ,
the assertion is as follows .
[ 3_4 ] for each integer @xmath28 , we define generating functions @xmath325 , @xmath326 , @xmath327 by @xmath328 then we have @xmath329 @xmath330 by definition , each generating function can be given as @xmath331 the lemma follows from these by a simple calculation using binomial theorem and we omit the details .
for the computation of the combinatorial part , we prepare the generating functions as follows .
@xmath332 @xmath333 @xmath334 @xmath335 @xmath336 where @xmath337 and @xmath338 are defined as in .
let @xmath339 then by definition we have @xmath340 the double shuffle relations of the combinatorial double eisenstein series are stated as [ 3_5 ] put @xmath341 then we have @xmath342 computations are parallel to those in @xcite , though tedious , and we omit the details .
the two lemmas and proposition 1 complete the proof of theorem 3 .
in this appendix we briefly recall the relation described in @xcite between the double eisenstein series and modular forms for @xmath199 .
the double eisenstein series for @xmath199 was first defined and studied in @xcite : @xmath343 its fourier series is given there as @xmath344 where @xmath345 , and @xmath346 by extending the definition in the case of non - absolute convergence using @xmath143-series , we showed that the double eisenstein series satisfy the double shuffle relations ( in the form described in @xcite ) , that the space of double eisenstein series contains the space of modular forms on @xmath199 , and made a connection to the period polynomial by looking at the imaginary parts of the @xmath143-expansions of @xmath347 .
specifically , the imaginary parts are given , like , by @xmath348 where @xmath349 is the @xmath350 matrix given by @xmath351 rather surprisingly , this contains exactly @xmath273 as a minor .
for example , @xmath352 precisely , the @xmath353-th row of @xmath273 is the @xmath354-st row of @xmath349 .
the right kernel of @xmath355 corresponds to the even period polynomials ( without constant term ) of weight @xmath57 for @xmath199 , an example being @xmath356 in the right kernel of @xmath357 and the corresponding period polynomial @xmath358 of weight @xmath359 . as in the corollary of theorem 5 , by looking at the constant term of the double eisenstein series and by using the connection to the period polynomial just mentioned , we obtain the upper bound of the dimension of the space of double zeta values : @xmath360 also , elements in the left kernel of @xmath355 produce expressions of modular forms in terms of double eisenstein series . by comparing the fourier coefficients , we obtain certain formulas for fourier coefficients of modular forms .
let us look at some examples in weight @xmath359 . as the simplest example
, take @xmath361 in the left kernel of @xmath357 .
this corresponds to the relation @xmath362 where @xmath363 is the famous cusp form of weight @xmath359 . comparing the coefficients of both sides , we obtain @xmath364 where @xmath365 incidentally , the ramanujan congruence @xmath366 is clearly seen from this . secondly take @xmath367 , which gives the relation @xmath368 and the formula @xmath369 as the third example , take @xmath370
this corresponds to the relation ( 8) in @xcite , and gives @xmath371 and @xmath372 we may take yet other vectors in the left kernel of @xmath357 ( the dimension is @xmath373 ) and may deduce similar kind of formulas for @xmath374 . +
* interestingly enough , the matrix @xmath355 appears when we write `` motivic '' double zeta values in terms of certain basis elements @xmath375 using coproduct structure described in f. brown s recent important papers @xcite .
one of the present authors has found the same relation between triple eisenstein series and motivic triple zeta values . or a variant ( minor matrix ) of @xmath355 appears in the work of s. baumard and l. schneps @xcite on a relation of double zeta values and period polynomials .
more precisely , the matrix @xmath376 ( we view a matrix as a union of row vectors and take the difference of sets ) is the matrix @xmath377 in @xcite .
the vector @xmath370 above is essentially the unique vector @xmath378 in the kernel of @xmath379 .
h. gangl , m. kaneko , and d. zagier , _
double zeta values and modular forms _ , automorphic forms and zeta functions , proceedings of the conference in memory of tsuneo arakawa , world scientific , ( 2006 ) , 71106 . | we study the double shuffle relations satisfied by the double zeta values of level 2 , and introduce the double eisenstein series of level 2 which satisfy the double shuffle relations .
we connect the double eisenstein series to modular forms of level 2 . |
that we are interested in an unknown @xmath7-by-@xmath2 matrix @xmath17 , thought to be either exactly or approximately of low rank , but we only observe a single noisy @xmath18-by-@xmath19 matrix @xmath20 , obeying @xmath21 .
the noise matrix @xmath22 has independent , identically distributed , zero - mean entries .
the matrix @xmath17 is a ( non - random ) parameter , and we wish to estimate it with some bound on the mean squared error ( mse ) .
the default estimation technique for our task is _ truncated svd _ ( tsvd ) @xcite : write @xmath23 for the singular value decomposition of the data matrix @xmath20 , where @xmath24 and @xmath25 , @xmath26 are the left and right singular vectors of @xmath20 corresponding to the singular value @xmath27 .
the tsvd estimator is @xmath28 where @xmath29 , assumed known , and @xmath30 .
being the best approximation of rank @xmath12 to the data in the least squares sense @xcite , and therefore the maximum likelihood estimator when @xmath22 has gaussian entries , the tsvd is arguably as ubiquitous in science and engineering as linear regression @xcite .
when the true rank @xmath31 of the signal @xmath17 is unknown , one might try to form an estimate @xmath32 and then apply the tsvd @xmath33 .
extensive literature has formed on methods to estimate @xmath31 : we point to the early @xcite ( in factor analysis and principal component analysis ) , the recent @xcite ( in our setting of singular value decomposition ) , and reference therein .
it is instructive to think about rank estimation ( using any method ) , followed by tsvd , simply as _ hard thresholding _ of the data singular values , where only components @xmath34 for which @xmath27 passes a specified threshold , are included in @xmath35 .
let @xmath36 denote the hard thresholding nonlinearity , and consider singular value hard thresholding ( svht ) @xmath37 in words , @xmath38 sets to @xmath1 any data singular value below @xmath39 .
matrix denoisers explicitly or implicitly based on hard thresholding of singular values have been proposed by many authors , including @xcite . as a common example of implicit svht denoising ,
consider the standard practice of estimating @xmath31 by plotting the singular values of @xmath20 in decreasing order , and looking for a `` large gap '' or `` elbow '' ( figure [ hist : fig ] , left panel ) .
when @xmath17 is exactly or approximately low - rank and the entries of @xmath22 are white noise of zero mean and unit variance , the empirical distribution of the singular values of the @xmath7-by-@xmath2 matrix @xmath21 forms a quarter - circle bulk whose edge lies approximately at @xmath40 , with @xmath41 @xcite .
only data singular values that are larger than the bulk edge are noticeable in the empirical distribution ( figure [ hist : fig ] , right plot ) .
since the singular value plot `` elbow '' is located at the bulk edge , the popular method of tsvd at the `` elbow '' is an approximation of _ bulk - edge hard thresholding _ , @xmath42 . sampled from the model @xmath43 , where @xmath44 , @xmath45 and @xmath46 elsewhere . in matlab
, a sample can be generated using the command y = svd(diag([[1.7 2.5 ] zeros(1,98)])+randn(100)/sqrt(100 ) ) .
left : the singular values of @xmath20 plotted in decreasing order , known in principal components analysis as a scree plot .
right : the singular values of @xmath20 in a histogram with @xmath47 bins . note the bulk edge approximately at @xmath48 , and note that the location of the top two singular values of @xmath20 is approximately @xmath49 ( @xmath50 ) , in agreement with lemma [ y - asy : lem ] . , width=355 ] let us measure the denoising performance of a denoiser @xmath35 at a signal
matrix @xmath17 using mean square error ( mse ) , @xmath51 the tsvd is an optimal rank-@xmath31 approximation of the data matrix @xmath20 , in mse .
but this does not necessarily mean that it is a good , or even reasonable , estimator to the signal matrix @xmath17 , which we wish to recover .
we may wonder : * _ question 1 .
_ assume that @xmath52 is unknown but small .
is there a singular value threshold @xmath39 so that svht @xmath38 successfully adapts to unknown rank and unknown noise level , and performs as well as tsvd would , had we known the true @xmath52 ? as we will see , it is convenient to represent the threshold as @xmath53 , where @xmath0 is a parameter typically between @xmath54 and @xmath55 .
recently , s. chatterjee @xcite proposed that one could have a single universal choice of @xmath0 ; that in a setting more general , but similar , to our setting , any @xmath56 would give near - optimal mse , in a qualitative sense ; and he specifically proposed @xmath57 , namely @xmath58 as a universal choice for svht , regardless of the shape @xmath59 of the matrix , and regardless of the underlying signal matrix @xmath17 or its rank . while the rule of @xcite was originally intended to be ` fairly good ' across many situations not reducible to the low - rank matrix in i.i.d noise model considered here , @xmath60 is a specific proposal , which prompts the following question : * _ question 2 .
_ is there really a single threshold parameter @xmath61 that provides good performance guarantees for mse ?
is that value @xmath62 ?
is it really independent of @xmath7 and @xmath2 ?
finally , note that singular value hard thresholding is just one strategy for matrix denoising .
it is not a - priori clear whether the whole idea of only ` keeping ' or ` killing ' empirical singular values based on their size makes sense .
could there exist a shrinkage rule @xmath63 , that more smoothly transitions from ` killing ' to ` keeping ' , which leads to a much better denoising scheme ?
we may wonder : * _ question 3 .
_ how does optimally tuned svht compare with the performance of the _ best possible shrinkage _ of singular values , at least in the worst - case mse sense ?
our main results imply that , in a certain asymptotic framework , there are simple and convincing answers to these questions .
following perry @xcite and shabalin and nobel @xcite , we adopt an asymptotic framework where the matrix grows while keeping the nonzero singular values of @xmath17 fixed , and the signal - to - noise ratio of those singular values stays constant with increasing @xmath2 . in this asymptotic framework , for a low - rank @xmath2-by-@xmath2 matrix observed in white noise of level @xmath3 , @xmath64 is the optimal location for the hard thresholding of singular values . for a non - square @xmath7-by-@xmath2 matrix with @xmath65 ,
the optimal location is @xmath66 where @xmath41 . the value @xmath67 is the _ optimal hard threshold coefficient _ for known @xmath3 .
it is given by formula below and tabulated for convenience in table [ lambdastar : tab ] .
( note added in proof : we found that p. perry s phd thesis @xcite proposes a threshold which can be shown to be equivalent to . )
our central observation is as follows .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ when a data singular value @xmath27 is too small , then its associated singular vectors @xmath68 are so noisy that the component @xmath69 should not included in @xmath35 . in our asymptotic framework , which models large , low - rank matrices observed in white noise , the cutoff below which @xmath27 is _ too small _ is exactly @xmath70 ( for square matrices ) . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * _ answer to question 1 : optimal svht dominates tsvd . _
optimally tuned svht @xmath71 is _ always _ at least as good as tsvd @xmath72 , in terms of amse ( theorem [ inadmis - tsvd : thm ] ) . unlike @xmath72
, the optimal svht @xmath71 does not require knowledge of @xmath29 .
in other words , it adapts to unknown low rank while giving uniformly equal or better performance . for square matrices ,
the tsvd provides a guarantee on worst - case amse that is @xmath73 times the guarantee provided by @xmath71 ( table [ mmx - square - boundedrank : tab ] ) . *
_ answer to question 2 : optimal svht dominates every other choice of hard threshold .
_ in terms of amse , optimally tuned svht @xmath71 is _ always _ at least as good as svht @xmath38 at any other fixed threshold @xmath74 ( theorem [ inadmis : thm ] ) .
it is the asymptotically minimax svht denoiser , over matrices of small bounded rank ( theorems [ mmx - square : thm ] and [ mmx - nonsquare : thm ] ) and over matrices of small bounded nuclear norm ( theorem [ mmx - l1:thm ] ) .
in particular , the parameter @xmath75 is noticeably worse . for square matrices
, @xmath60 provides a guarantee for worst - case amse that is @xmath76 times the guarantee provided by @xmath71 ( table [ mmx - square - boundedrank : tab ] ) . * _ answer to question 3 .
optimal svht compares adequately to the optimal shrinker .
_ optimally tuned svht @xmath71 provides a guarantee on worst - case asymptotic mse that is @xmath77 times ( for square matrices ) the best possible guarantee achievable by _ any _ shrinkage of data singular values ( table [ mmx - square - boundedrank : tab ] ) .
these are all rigorous results , within a specific asymptotic framework , which prescribes a certain scaling of the noise level , the matrix size , and the signal - to - noise ratio as @xmath2 grows .
but does amse predict actual mse in finite - sized problems ? in section [ msevsamse : subsec ] we show finite-@xmath2 simulations demonstrating the effectiveness of these results even at rather small problem sizes . in high signal - to - noise , all denoisers considered here
perform roughly the same , and in particular the classical tsvd is a valid choice in that regime .
however , in low and moderate snr , the performance gain of optimally tuned svht is substantial , and can offer @xmath78 decrease in amse . for a low - rank @xmath2-by-@xmath2 matrix observed in white noise of _ unknown _ level , one can use the data to obtain an approximation of the optimal location @xmath79 . define @xmath80 where @xmath6 is the median singular value of the data matrix @xmath20 .
the notation @xmath81 is meant to emphasize that this is not a fixed threshold chosen a - priori , but rather a data - dependent threshold . for a non - square @xmath7-by-@xmath2 matrix with @xmath65 , the approximate optimal location when @xmath3 is unknown is @xmath82
the optimal hard threshold coefficient for unknown @xmath3 , denoted by @xmath83 , is not available as an analytic formula , but can easily be evaluated numerically .
we provide a matlab script for this purpose @xcite ; the underlying derivation appears in section [ sigmahat : subsec ] below .
some values of @xmath83 are provided in table [ lambdastar - hat : tab ] . when a high - precision value of @xmath83 can not be computed
, one can use the approximation @xmath84 the optimal svht for unknown noise level , @xmath85 , is very simple to implement and does not require any tuning parameters .
the denoised matrix @xmath86 can be computed using just a few code lines in a high - level scripting language .
for example , in matlab : .... beta = size(y,1 ) / size(y,2 ) ; omega = 0.56*beta^3 - 0.95*beta^2 + ... 1.82*beta + 1.43 ; [ u d v ] = svd(y ) ; y = diag(y ) ; y ( y < ( omega * median(y ) ) = 0 ; xhat = u * diag(y ) * v ' ; .... here we have used the approximation .
we recommend , whenever possible , to use a function omega(beta ) , such as the one we provide in the code supplement @xcite , to compute the coefficient @xmath83 to high precision .
in our asymptotic framework , @xmath79 and @xmath81 enjoy exactly the same optimality properties .
this means that @xmath85 adapts to unknown low rank _ and _ to unknown noise level .
empirical evidence suggest that their performance for finite @xmath2 is similar . as a result ,
the answers we provide above hold for the threshold @xmath81 when the noise level is unknown , just as they hold for the threshold @xmath79 when the noise level is known .
column vectors are denoted by boldface lowercase letters , such as @xmath87 , their transpose is @xmath88 and their @xmath89-th coordinate is @xmath90 .
the euclidean inner product and norm on vectors are denoted by @xmath91 and @xmath92 , respectively .
matrices are denotes by uppercase letters , such as @xmath17 , its transpose is @xmath93 and their @xmath94-th entry is @xmath95 .
@xmath96 denotes the space of real @xmath7-by-@xmath2 matrices , @xmath97 denotes the hilbert-schmidt inner product , and @xmath98 denotes the corresponding frobenius norm on @xmath96 . for simplicity
we only consider @xmath99 .
we denote matrix denoisers , or estimators , by @xmath100 . the symbols @xmath101 and
@xmath102 denote almost sure convergence and equality of a.s .
limits , respectively . with the exception of tsvd ,
when @xmath3 is known , all the denoisers we discuss operate by shrinkage of data singular values , namely are of the form @xmath103 where @xmath20 is given by and @xmath63 is some univariate shrinkage rule . as we will see , in the general model @xmath21 , the noise level in the singular values of @xmath20 is @xmath104 . instead of specifying a different shrinkage rule that depends on the matrix size @xmath2 , we calibrate our shrinkage rules to the `` natural '' model @xmath105 . in this convention ,
shrinkage rules stay the same for every value of @xmath2 , and we conveniently abuse notation by writing @xmath35 as in for any @xmath100 , keeping @xmath7 and @xmath2 implicit . to apply any denoiser @xmath35 below to data from the general model @xmath21 , use the denoiser @xmath106 for example , to apply the svht @xmath107 to data sampled from the model @xmath21 , use @xmath108 , with @xmath109 throughout the text
, we use @xmath110 to denote svht calibrated for noise level @xmath111 and @xmath112 to denote svht calibrated for a specific general model @xmath21 . to translate the amse of any denoiser @xmath35 , calibrated for noise level @xmath111 , to an approximate mse of the corresponding denoiser @xmath113 , calibrated for a model @xmath21
, we use the identity lcl + & & n^2|| ( x/())+z/)-x/ ( ) ||_f^2 .
below , we spell out this translation of amse where appropriate . in this paper , we consider a sequence of increasingly larger denoising problems @xmath114 , with @xmath115 , satisfying the following assumptions : 1 . _ invariant white noise : _ the entries of @xmath116 are i.i.d samples from a distribution with zero mean , unit variance and finite fourth moment . to simplify the formal statement of our results , we assume that this distribution is _ orthogonally invariant _ in the sense that @xmath116 follows the same distribution as @xmath117 , for every orthogonal @xmath118 and @xmath119 .
this is the case , for example , when the entries of @xmath116 are gaussian . in section [ general_noise : sec ]
we revisit this restriction and discuss general ( not necessarily invariant ) white noise .
fixed signal column span @xmath120 : _ let the rank @xmath121 be fixed and choose a vector @xmath122 with coordinates @xmath123 such that @xmath124 .
assume that for all @xmath2 , @xmath125 is an arbitrary and nonzero signal singular values @xmath126 are shared by all matrices @xmath127 , the signal left and right singular vectors @xmath128 and @xmath129 are unknown and arbitrary . ] singular value decomposition of @xmath127 , where @xmath130 and @xmath131 .
_ asymptotic aspect ratio @xmath132 : _ the sequence @xmath133 is such that @xmath134 . to simplify our formulas , we assume that @xmath135 .
let @xmath35 be any singular value shrinkage denoiser calibrated , as discussed above , for noise level @xmath111 .
define the asymptotic mse ( amse ) of an @xmath35 at a signal @xmath136 by the ( almost sure ) limit . ]
@xmath137 adopting the asymptotic framework above , we seek singular value thresholding rules @xmath110 that minimize the amse @xmath138 .
as we will see , in this framework there are simple , satisfying answers to the questions posed in the introduction .
define the _ optimal hard threshold for singular values _ for @xmath2-by-@xmath2 square matrices by @xmath139 more generally , define the optimal threshold for @xmath18-by-@xmath19 matrices with @xmath140 by @xmath141 some values of @xmath67 are provided in table [ lambdastar : tab ] . .some optimal hard threshold coefficients @xmath67 from . for @xmath7-by-@xmath2 matrix in known noise level @xmath3 ( with @xmath142 ) ,
the optimal svht denoiser @xmath71 sets to zero all data singular values below the threshold @xmath143 . [ cols=">,<,>,<",options="header " , ]
figure [ lambda : fig ] shows the optimal threshold @xmath67 over @xmath132 .
the edge of the quarter circle bulk @xmath144 , the hard threshold that best emulates tsvd in our setting , is shown for comparison . in the null case @xmath145 , the largest data singular value is located asymptotically exactly at the bulk edge , @xmath146 .
it might seem that just above the bulk edge is a natural place to set a threshold , since anything smaller could be the product of a pure noise situation .
however , for @xmath147 , the optimal hard threshold @xmath148 is 15 - 20% larger than the bulk edge ; as @xmath149 , it grows about 40% larger . inspecting the proof of theorem [ inadmis : thm ] and particularly the expression for amse of svht ( lemma [ amse : lem ] )
, one finds the reason : one component of the amse is due to the angle between the signal singular vectors and the data singular vectors .
this angle converges to a nonzero value as @xmath150 ( given explicitly in lemma [ inner - asy : lem ] ) which grows as snr decreases . when some data singular value @xmath27 is too close to the bulk , its corresponding singular vectors are too badly rotated , and the rank - one matrix @xmath151 it contributes to the denoiser hurts the amse more than it helps .
for example , for square matrices @xmath152 , this situation is most acute when the signal singular value is just barely larger than @xmath153 , causing the corresponding data singular value @xmath27 to be just barely larger than the bulk edge , which for square matrices is located at @xmath48 .
a svht denoiser thresholding just above the bulk edge would include the component @xmath154 , incurring an amse about 5 times larger than the amse incurred by excluding @xmath27 from the reconstruction .
the optimal threshold @xmath67 keeps such singular values out of the picture ; this is why it is necessarily larger than the bulk edge .
the precise value of @xmath67 is the precise point at which it becomes advantageous to include the rank - one contribution of a singular value @xmath27 in the reconstruction . from against @xmath132 .
also shown are bulk - edge @xmath144 , which is the hard threshold corresponding to tsvd in our setting , and the usvt threshold @xmath62 from @xcite .
, width=326 ] as mentioned in the introduction , s. chatterjee has recently discussed svht in a broad class of situations @xcite . translating his much broader discussion to the confines of the present context , he observed that _ any _ @xmath56 can serve as a _ universal hard threshold for singular values _ ( usvt ) , offering fairly good performance regardless of the matrix shape @xmath59 and the underlying signal matrix @xmath17 .
the author makes the specific recommendation @xmath57 and writes : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the algorithm manages to cut off the singular values at the ` correct ' level , depending on the structure of the unknown parameter matrix .
the adaptiveness of the usvt threshold is somewhat similar in spirit to that of the sureshrink algorithm of donoho and johnstone .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ keeping in mind that the scope of @xcite is much broader than the one considered here , we would like to evaluate this proposal , in the setting of low rank matrix in white noise , and specifically in our asymptotic framework .
figure [ lambda : fig ] includes the value @xmath62 : indeed , this threshold is larger than the bulk edge , for any @xmath156 , so chatterjee s @xmath155 rule asymptotically set to zero all singular values which _
could _ arise due to an underlying noise - only situation .
when @xmath157 , the @xmath155 rule sometimes `` kills '' singular values that the optimal threshold deems good enough for keeping , and when @xmath158 , the @xmath155 rule sometimes `` keeps '' singular values that did in fact arise from signal , but are so close to the bulk that the optimal threshold declares them unusable . for @xmath152 ,
the guarantee on worst - case amse obtained by using @xmath57 over matrices of rank @xmath12 is about @xmath159 , roughly 140% larger than the guarantee obtained by using the minimax threshold @xmath160 ( see figure [ amse : fig ] ) . for square matrices ,
the regret for preferring usvt to optimally - tuned svht can be substantial : in low snr ( @xmath161 ) , using the threshold @xmath57 incurs roughly twice the amse of the minimax threshold @xmath9 .
we note that unlike the optimally tuned svht @xmath162 , the usvt @xmath155 does not take into account the shape factor @xmath132 , namely the ratio of number of rows to number of columns of the matrix in question .
a comparison of worst - case amse between the fixed threshold choice @xmath57 and the optimal hard threshold @xmath163 is shown in figure [ sourav - mmx - lf : fig ] .
the two curves intersect at @xmath164 , where the optimal threshold is approximately @xmath62 .
, for two choices of hard threshold : @xmath57 from @xcite , and optimal threshold @xmath61 from .
, width=326 ] one might argue that @xcite proposed @xmath62 based on its mse performance over classes of matrices bounded in nuclear norm .
but also for that purpose , @xmath62 is noticeably outperformed by @xmath148 .
arguing as in theorem [ mmx - l1:thm ] we obtain , in the square case : @xmath165 the coefficient @xmath166 is about 110% larger than the best coefficient achievable by svht , namely @xmath167 in .
one should keep in mind that usvt is applicable for a wide range of noise models , e.g. in stochastic block models .
@xcite is the first , to the best of out knowledge , to suggest that a matrix denoising procedure as simple as svht could have universal optimality properties . in our asymptotic framework of low - rank matrices in white noise , the 2.02 threshold performs fairly well in amse , except for very small values of @xmath132 ( figure [ amse : fig ] ) ; but one often gets a substantial amse improvement by switching to the rule we recommend . since our recommendation dominates in amse , there is no downside to making this switch
i.e. there is no configuration of signal singular values @xmath168 which could make one regret this switch .
setting additional notation required in the proofs , let @xmath169 be a sequence of signal matrices in our asymptotic framework , so that @xmath170 ( resp .
@xmath171 ) is the left ( resp .
right ) singular vector corresponding to the singular value @xmath172 , namely , @xmath89-th column of @xmath128 ( resp .
@xmath129 ) in .
similarly , let @xmath173 be a corresponding sequence of observed matrices in our framework , and write @xmath174 so that @xmath175 ( resp .
@xmath176 ) is the left ( resp .
right ) singular vector corresponding to the singular value @xmath177 .
( note that @xmath178 and @xmath179 are unknown , arbitrary , non - random vectors . )
our main results depend on lemma [ amse : lem ] , a formula for the amse of svht .
this formula in turn depends on lemma [ y - asy : lem ] and lemma [ inner - asy : lem ] .
both follow from recent key results due to @xcite .
[ y - asy : lem ] * asymptotic data singular values . * for @xmath180 , @xmath181 [ inner - asy : lem ] * asymptotic angle between signal and data singular vectors .
* let @xmath182 and assume that @xmath172 has degeneracy @xmath183 , namely , there are exactly @xmath183 entries of @xmath136 equal to @xmath172 . if @xmath184 , we have @xmath185 and , a slightly different formula , @xmath186 if however @xmath187 , then we have @xmath188 to appeal to these results , we need to show that our asymptotic framework satisfies the assumptions of @xcite . by @xcite the limiting law of the singular values of @xmath189 is the quarter - circle density @xmath190}}(x)\,;\end{aligned}\ ] ] by @xcite , @xmath191 ; by @xcite , @xmath192 .
this satisfies assumptions 2.1 , 2.2 and 2.3 of @xcite , respectively
. formulas , and , as seen in ( * ? ? ?
* example 3.1 ) , depend only on the shape of the limiting distribution and not on any gaussian assumptions . using lemma [ y - asy : lem ] and lemma [ inner - asy : lem ] , we can calculate the amse of the hard thresholding estimator @xmath110 , for given threshold @xmath0 , at a matrix of specific aspect ratio @xmath132 and signal singular values @xmath136 : [ amse : lem ] * amse of singular value hard thresholding . * fix @xmath121 and @xmath122 .
let @xmath193 and @xmath194 be matrix sequences in our asymptotic framework , and let @xmath195 .
then @xmath196 where lcl [ m : eq ] + & & ( x+)(x + ) - ( x^2- ) & x x _ * ( ) + x^2 & x < x _ * ( ) and @xmath197 is given by eq . .
figure [ amse : fig ] shows the amse of lemma [ amse : lem ] , in square case @xmath152 and nonsquare cases @xmath198 , @xmath199 and @xmath200 . by definition ,
@xmath201 where @xmath202 .
observe that llcl [ finite - dif : eq ] + & & & @xmath203 = + & & & @xmath204 + + & & & @xmath205 -2@xmath206= + & & & _ i=1^m_n _ h(y_n , i;)^2 + _ i=1^rx_i^2 - + & & & 2_i , j=1^rx_i _ h(y_n , j ; ) @xmath207 = + & & & _ i = r+1^m_n _ h(y_n , i;)^2 + _ i=1^r . since @xmath208
, the leftmost term above converges almost surely to zero . when @xmath209 , by lemma [ y - asy : lem ] and lemma [ inner - asy : lem ] , only the term @xmath172 survives and eq .
assume now that @xmath184 .
we now consider the a.s . limiting value of each of the remaining terms in . for the term @xmath210 , by lemma [ y - asy : lem ] , for @xmath211 we have lcl + & & ( x_i+)(x_i + ) & ( x_i+)(x_i + ) ^2 + 0 & ( x_i+)(x_i + ) < ^2
turning to the rightmost term of , by lemma [ inner - asy : lem ] , for @xmath212 we find that it equals lcl + & & & x_i = x_j + 0 & x_ix_j where @xmath213 .
furthermore , since for all @xmath214 and @xmath215 we have @xmath216 , we find that for @xmath211 , lcl + & & _ j=1^r(y_n , j ; ) @xmath217 @xmath218 . for the rightmost term of we conclude that lcl + & & _ 1jr : x_j = x_i _ n = + & & & ( x_i+)(x_i + ) ^2 + 0 & ( x_i+)(x_i + ) < ^2 , where we have used lemma [ y - asy : lem ] again . collecting the terms ,
we find for the limiting value of that @xmath219 where @xmath220 is given by as required .
for the tsvd , the same argument gives : [ amse - tsvd : lem ] * amse of tsvd .
* fix @xmath121 and @xmath122 .
let @xmath193 and @xmath194 be matrix sequences in our asymptotic framework , and let @xmath195 .
then @xmath221 where lcl [ m_tsvd : eq ] + & & ( x+)(x + ) - ( x^2- ) & x ^1/4 + ( 1+)^2 + x^2 & x ^1/4 .
we now to turn to prove our main results .
let @xmath222 where @xmath67 is defined in and @xmath197 is defined in
. then @xmath223 it follows that for all @xmath224 and @xmath195 , lcl + & = & \ { x^2 , ( x+)(x+ ) - ( x^2- ) } + & & m(_,x ) and the theorem follows from eq . .
figure [ lf : fig ] provides a visual explanation of this proof for the square ( @xmath152 ) case . for @xmath225 , by lemma [ amse : lem ] and lemma
[ amse - tsvd : lem ] we have @xmath226 for @xmath227 , by lemma [ amse - tsvd : lem ] and theorem [ inadmis : thm ] we have @xmath228 theorem [ mmx - square : thm ] is a special case of theorem [ mmx - nonsquare : thm ] . by , it is enough to consider the univariate function @xmath229 defined in .
the theorem follows from lemma [ amse : lem ] using the following simple observation .
let @xmath156 and @xmath230 .
denote by @xmath197 the unique positive solution to the equation @xmath231 .
let @xmath61 be the unique solution to the equation in @xmath0 @xmath232 then for the function @xmath220 defined in , we have @xmath233 note that the least favorable situation occurs when @xmath234 , and that @xmath197 is precisely the value of @xmath168 for which the corresponding limiting data singular value satisfies @xmath235 . in other words ,
the least favorable situation occurs when the data singular values all coincide with each other and with the chosen hard threshold .
let @xmath236 denote the empirical cumulative distribution function ( cdf ) of the squared singular values of @xmath237 .
write @xmath238 where @xmath239 is a functional which takes as argument the cdf and delivers the median of that cdf . under our asymptotic framework ,
almost surely , @xmath236 converges weakly to a limiting distribution , @xmath240 , the cdf of the marenko - pastur distribution with shape parameter @xmath132 @xcite .
this distribution has a positive density throughout its support , in particular at its median .
the median functional is continuous for weak convergence at @xmath241 , and hence , almost surely , @xmath242 it follows that , @xmath243
our results were formally stated for a sequence of models of the form @xmath21 , where @xmath17 is a non - random matrix to be estimated , and the entries of @xmath22 are i.i.d samples from a distribution that is orthogonally invariant ( in the sense that the matrix @xmath22 follows the same distribution as @xmath244 , for any orthogonal @xmath245 and @xmath119 ) . while gaussian noise is orthogonally invariant , many common distributions , which one could consider to model white observation noise , are not .
one attractive feature of the discussion on optimal choice of singular value hard threshold , presented above , is that the amse @xmath246 only depends on the signal matrix @xmath17 through its rank , or more specifically , through its nonzero singular values @xmath136 . if the distribution of @xmath22 is not orthogonally invariant , mse ( or amse ) losses this property and depends on properties of @xmath17 other than its rank .
this point is discussed extensively in @xcite . in general white noise , which is not necessarily orthogonally invariant
, one can still allow mse to depend on @xmath17 only through its singular values by placing a prior distribution on @xmath17 and shifting to a model where it is a random , instead of a fixed , matrix .
specifically , consider an alternative asymptotic framework to the one in section [ framework : subsec ] , in which the sequence denoising problems @xmath114 satisfies the following assumptions : 1 .
_ general white noise : _ the entries of @xmath116 are i.i.d samples from a distribution with zero mean , unit variance and finite fourth moment .
2 . _ fixed signal column span and uniformly distributed signal singular vectors : _ let the rank @xmath121 be fixed and choose a vector @xmath122 with coordinates @xmath123 .
assume that for all @xmath2 , @xmath125 is a singular value decomposition of @xmath127 , where @xmath128 and @xmath129 are uniformly distributed random orthogonal matrices .
formally , @xmath128 and @xmath129 are sampled from the haar distribution on the @xmath7-by-@xmath7 and @xmath2-by-@xmath2 orthogonal group , respectively .
asymptotic aspect ratio @xmath132 : _ the sequence @xmath133 is such that @xmath134 .
the second assumption above implies that @xmath127 a `` generic '' choice of matrix with nonzero singular values @xmath136 , or equivalently , a generic choice of coordinate systems in which the linear operator corresponding to @xmath17 is expressed .
the results of @xcite , which we have used , hold in this case as well .
it follows that lemma [ amse : lem ] and lemma [ amse - tsvd : lem ] , and consequently all our main results , hold under this alternative framework .
in short , in general white noise , all our results hold if one is willing to only specify the signal singular values , rather than the signal matrix , and consider a `` generic '' signal matrix with these singular values .
we have calculated the exact optimal threshold @xmath79 in a certain asymptotic framework .
the practical significance of our results hinges on the validity of the amse as an approximation to mse , for values of @xmath247 and error distributions encountered in practice .
this in turn depends on the simultaneous convergence of three terms : * convergence of the top data singular values @xmath177 ( @xmath248 ) to the limit in lemma [ y - asy : lem ] , * convergence of the angle between the top data singular vectors @xmath249 and their respective signal singular vectors to the limit in lemma [ inner - asy : lem ] , and * convergence of the rest of the data singular values @xmath177 ( @xmath250 to the interval @xmath251 $ ] .
analysis of each of these terms for specific error distributions is beyond our current scope .
figure [ 4a : fig ] contains a few sample comparisons of amse and empirical mse we have performed .
the matrix sizes and number of monte carlo draws are small enough to demonstrate that amse is a reasonable approximation even for relatively small low - rank matrices . as convergence of the empirical spectrum to its limit
is known to depend on moments of the underlying distributions , we include results for different error distributions .
amse is found to be a useful proxy to mse even in small matrix sizes .
amse of svht was found to be inaccurate when : ( i ) the rank fraction is nontrivial ( e.g @xmath252 , @xmath253 shown at the bottom of figure [ 4a : fig ] ) ; ( ii ) the threshold @xmath0 is very close to the approximate bulk edge @xmath254 . in case ( i ) ,
interaction effects between singular values , which are ignored in our asymptotic framework , start to have non - negligible effect . in case ( ii ) , where the discontinuity of the svht nonlinearity is placed close to the bulk edge , the distribution of the largest `` non - signal '' singular value @xmath255 , which is known in some cases to be asymptotically a tracy - widom distribution @xcite , becomes important .
indeed , some data singular values from the bulk manage to pass the threshold @xmath0 and cause their singular vectors to be included in the estimator @xmath110 .
our derivation of amse assumed however that no such singular vectors are included in @xmath110 , since @xmath256 .
note however that the main recommendation of this paper is that one should _ not _ threshold at or near he bulk edge , as explained in detail above .
therefore , from a practical perspective , the inaccuracy of amse for svht with @xmath0 near the bulk edge is slightly irrelevant .
the asymptotic framework considered here is perhaps the simplest nontrivial model for matrix denoising .
it allows one to calculate , in amse , basically any quantity of interest , for any denoiser of interest .
the fundamental elements of matrix denoising in white noise , which underly more complicated models , are present yet understandable and quantifiable .
for example , the amse of any denoiser based on singular value shrinkage contains a component due to noise contamination in the data singular vectors , and this component determines a fundamental lower bound on amse . we conjecture that results calculated in this model , which are not attached to a specific assumption on rank ( e.g , the constants in table [ mmx - square - l1:tab ] , which determine the minimax amse over nuclear norm balls ) remain essentially correct in more complicated models .
the decision - theoretic landscape as it appears through the naive prism of our asymptotic framework is extremely simple : there is a unique admissible hard thresholding rule , and moreover a unique admissible shrinkage rule , for singular values .
this is of course quite different from the situation encountered , for example , in estimating normal means .
the reason is the extreme simplicity of our model .
for example , we have replaced the data singular values , which are random for finite matrix size , with their almost sure limits , and in effect neglected their random fluctuations around these limits .
these fluctuations are now well understood ( see for example @xcite ) .
we have ignored this structure . however , including these second - order terms in the asymptotic distributions is only likely to achieve second - order improvements in mse over our suggested optimal truncation threshold .
in the code supplement @xcite we offer a matlab software library that includes : 1 . a function that calculates the optimal shrinkage coefficient in known or unknown noise level .
scripts that generate each of the figures in this paper .
3 . a script that generates figures similar to figure [ 4a : fig ] , which compare amse to mse in various situations .
the authors would like to thank andrea montanari for pointing to the work of shabalin and nobel , drew nobel and sourav chatterjee for helpful discussions , art owen for pointing to the work of perry , and the anonymous referees for their useful suggestions .
this work was partially supported by nsf dms 0906812 ( arra ) .
mg was partially supported by a william r. and sara hart kimball stanford graduate fellowship .
d. l. donoho and m. gavish , `` code supplement to ` the optimal hard threshold for singular values is @xmath9 ' , '' http://purl.stanford.edu/vg705qn9070 , 2014 , accessed 27 march 2014 .
[ online ] .
available : http://purl.stanford.edu/vg705qn9070 o. alter , p. brown , and d. botstein , `` singular value decomposition for genome - wide expression data processing and modeling , '' _ proceedings of the national academy of sciences _ , vol . 97 , no . 18 , pp . 1010110106 , aug . 2000 .
t. d. lagerlund , f. w. sharbrough , and n. e. busacker , `` spatial filtering of multichannel electroencephalographic recordings through principal component analysis by singular value decomposition . ''
_ journal of clinical neurophysiology : official publication of the american electroencephalographic society _ , vol .
14 , no . 1 ,
7382 , jan . 1997 . a. l. price , n. j. patterson , r. m. plenge , m. e. weinblatt , n. a. shadick , and d. reich , `` principal components analysis corrects for stratification in genome - wide association studies . '' _ nature genetics
_ , vol .
38 , no . 8 , pp .
9049 , aug .
2006 .
s. wold , `` cross - validatory estimation of the number of components in factor and principal components components models , '' _ technometrics _ , vol .
20 , no . 4 , pp .
397405 , 1978 .
[ online ] .
available : http://www.tandfonline.com/doi/pdf/10.1080/00401706.1978.10489693 a. b. owen and p. o. perry , `` bi - cross - validation of the svd and the nonnegative matrix factorization , '' _ the annals of applied statistics _
, vol . 3 , no . 2 , pp .
564594 , jun . 2009 .
[ online ] .
available : http://projecteuclid.org/euclid.aoas/1245676186 d. achlioptas and f. mcsherry , `` fast computation of low rank matrix approximations , '' in _ proceedings of the thirty - third annual acm symposium on theory of computing _ , 2001 , pp .
[ online ] .
available : http://dl.acm.org/citation.cfm?id=380858 y. azar , a. fiat , a. r. karlin , f. mcsherry , and j. saia , `` spectral analysis of data , '' in _ proceedings of the thirty - third annual acm symposium on theory of computing _ , 2001 , pp .
[ online ] .
available : http://dl.acm.org/citation.cfm?id=380859 p. j. bickel and e. levina , `` covariance regularization by thresholding , '' _ the annals of statistics _
36 , no . 6 , pp .
25772604 , dec .
[ online ] .
available : http://projecteuclid.org/euclid.aos/1231165180 r. h. keshavan , a. montanari , and s. oh , `` matrix completion from a few entries , '' _ ieee transcations on information theory _
56 , no . 6 , pp.2980 - 2998 , 2010 .
[ online ] .
available : http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5466511 d. l. donoho , m. gavish and i. m. johnstone , `` optimal shrinkage of eigenvalues in the spiked covariance model , '' _ stanford university statistics department technical report 2013 - 10 _ , 2013 .
[ online ] .
available : http://arxiv.org/abs/1311.0851 f. benaych - georges and r. r. nadakuditi , `` the singular values and vectors of low rank perturbations of large rectangular random matrices , '' _ journal of multivariate analysis _
120135 , oct . 2012 .
y. q. yin , z. d. bai , and p. r. krishnaiah , `` on the limit of the largest eigenvalue of the large dimensional sample covariance matrix , '' _ probability theory and related fields _ , vol .
78 , pp . 509-521 , 1988 .
cai , e. j. cands , and z. shen , `` a singular value thresholding algorithm for matrix completion , '' _ siam journal on optimization _ , vol .
20 , no . 4 , p. 1956 , 2008 .
[ online ] .
available : http://arxiv.org/abs/0810.3286 e. j. cands , c. a. sing - long , and j. d. trzasko , `` unbiased risk estimates for singular value thresholding and spectral estimators , '' _ ieee transactions on signal processing _ , vol .
61 , no .
19 , pp . 46434657 , 2012 .
d. l. donoho and m. gavish , `` minimax risk of matrix denoising by singular value thresholding , '' _ stanford university statistics department technical report 2013 - 03 _ , 2013 .
[ online ] .
available : http://arxiv.org/abs/1304.2085 s. kritchman and b. nadler , `` non - parametric detection of the number of signals : hypothesis testing and random matrix theory , '' _ signal processing , ieee transactions _ , vol .
57 , no .
10 , pp . 39303941 , 2009 .
z. bai and j .- f .
yao , `` central limit theorems for eigenvalues in a spiked population model , '' _ annales de linstitut henri poincare ( b ) probability and statistics _
44 , no . 3 , pp .
447474 , jun . 2008 .
matan gavish received the dual b.sc .
degree in mathematics and physics from tel aviv university ( tau ) in 2006 and the m.sc .
degree in mathematics from the hebrew university of jerusalem in 2008 .
he is currently a doctoral student in statistics at stanford university , in collaboration with the yale university program in applied mathematics .
his research interests include applied harmonic analysis , high - dimensional statistics and computing .
he was in the adi lautman interdisciplinary program for outstanding students at tau from 2002 to 2006 and held a william r. and sara hart kimball stanford graduate fellowship from 2009 to 2012 .
david l. donoho is a professor at stanford university .
his research interests include computational harmonic analysis , high - dimensional geometry , and mathematical statistics .
dr . donoho received the ph.d .
degree in statistics from harvard university , and holds honorary degrees from university of chicago and ecole polytechnique federale de lausanne .
he is a member of the american academy of arts and sciences and the us national academy of sciences , and a foreign associate of the french acadmie des sciences . | we consider recovery of low - rank matrices from noisy data by hard thresholding of singular values , in which empirical singular values below a threshold @xmath0 are set to @xmath1 .
we study the asymptotic mse ( amse ) in a framework where the matrix size is large compared to the rank of the matrix to be recovered , and the signal - to - noise ratio of the low - rank piece stays constant .
the amse - optimal choice of hard threshold , in the case of @xmath2-by-@xmath2 matrix in white noise of level @xmath3 , is simply @xmath4 when @xmath3 is known , or simply @xmath5 when @xmath3 is unknown , where @xmath6 is the median empirical singular value . for nonsquare @xmath7 by @xmath2 matrices with @xmath8 the thresholding coefficients @xmath9 and @xmath10
are replaced with different provided constants that depend on @xmath11 .
asymptotically , this thresholding rule adapts to unknown rank and unknown noise level in an optimal manner : it is _ always _
better than hard thresholding at any other value , and is _ always _
better than ideal truncated svd ( tsvd ) , which truncates at the true rank of the low - rank matrix we are trying to recover .
hard thresholding at the recommended value to recover an @xmath2-by-@xmath2 matrix of rank @xmath12 guarantees an amse at most @xmath13 . in comparison ,
the guarantees provided by tsvd , optimally tuned singular value soft thresholding and the best guarantee achievable by any shrinkage of the data singular values are @xmath14 , @xmath15 , and @xmath16 , respectively .
the recommended value for hard threshold also offers , among hard thresholds , the best possible amse guarantees for recovering matrices with bounded nuclear norm .
empirical evidence suggests that performance improvement over tsvd and other popular shrinkage rules can be substantial , for different noise distributions , even in relatively small @xmath2 . singular values shrinkage , optimal threshold , low - rank matrix denoising , unique admissible , scree plot elbow truncation , quarter circle law , bulk edge . |
the @xmath8cdm paradigm , in many respects , has proven to been an immensely successful cosmological model .
@xmath8cdm is based on a cosmological constant plus `` cold '' dark matter , i.e. dark matter particles whose streaming velocities are negligible for most astrophysical considerations . on large scales ,
the _ planck _
mission has just delivered another ringing endorsement of this model with its first - year cosmology results @xcite . on scales below a few ( comoving ) mpc , however , the matter power spectrum is still difficult to probe , and it has been repeatedly suggested that dark matter is perhaps `` warm '' , with a free - streaming length that affects the properties of low - mass ( dwarf ) galaxies .
warm dark matter ( wdm ) could alleviate the apparent difficulties of @xmath8cdm models in reproducing some observations related to the matter power spectrum on scales of a few mpc and below .
the most notable possible tensions under @xmath8cdm are : the excess of the number of galactic satellites , the cuspiness and high ( phase space ) density of galactic cores , the luminosities of the milky way s satellites and the properties of galaxies filling voids ( e.g. @xcite ) . the main effect of the larger velocities of wdm particles , and the resulting significant free - streaming length , would be to suppress structures on mpc scales and below .
the last few years have seen a re - intensified discussion of this possibility , particularly in light of improvements in numerical models and observations of the mass and internal structure of local group satellites @xcite . it has been suggested that a free - streaming length corresponding to that of a thermal relic wdm particle with a mass of 1 - 2 kev ( and in some cases as low as 0.5 kev ) provides better agreement between the most recent data and numerical simulations @xcite .
the difficulties associated with the cold dark matter paradigm , however , arise on scales where the matter spectrum is highly non - linear at @xmath11 , and where very uncertain baryonic physics is known to play an important role @xcite .
the lyman-@xmath0 absorption produced by intergalactic neutral hydrogen in the spectra of distant quasars ( qsos)the so called `` lyman-@xmath0 forest''provides a powerful alternative tool for constraining dark matter properties , particularly the free - streaming of dark matter particles on the scales in question . the lyman-@xmath0 forest probes the matter power spectrum in the mildly non - linear regime over a large range of redshifts ( @xmath12 in ground - based data ) down to the small scales of interest ( @xmath13 mpc ) @xcite . on large scales
the sdss - iii boss collaboration has recently measured the baryonic acoustic oscillations ( baos ) scale in the 3d correlation function of the lyman-@xmath0 forest in @xmath14 50,000 qsos at @xmath15 @xcite .
these findings further emphasize the value of the lyman-@xmath0 forest as a tracer of cosmological large - scale structure .
constraints on the matter power spectrum from lyman-@xmath0 forest data on small scales are only limited by the thermal cut - off in the flux power spectrum introduced by pressure and thermal motions of baryons in the photo - ionized intergalactic medium ( igm ) .
the igm has a characteristic temperature of @xmath16 . while not trivial , modeling the relevant physics with numerical simulations is reasonably straightforward
; the power spectrum at the relevant redshifts ( @xmath17 ) and scales is only mildly non - linear and stellar feedback effects are much less important than at lower redshifts @xcite .
the basic property of wdm , which impacts on both large scale structure formation and the internal structure of dark matter haloes and the galaxies they are hosting , is the significant `` thermal '' velocities of the wdm particles ( see @xcite ) .
the resulting `` free - streaming '' eliminates density fluctuations on scales below a characteristic comoving wavenumber : @xmath18 and leads to a very distinctive cut - off in the matter power spectrum at a corresponding scale .
for example , the wavenumber at which the linear wdm suppression reaches 50% in terms of matter power , @xmath19 , w.r.t .
the @xmath8cdm case can be approximated as : @xmath20 where this equation uses the numerical results of ref .
@xcite . for standard thermal relics ,
the shape of the cut - off is therefore well characterized in the linear regime and there is an unambiguous relation between the mass of the thermal relic wdm particle and a well - defined free - streaming length ( e.g. @xcite ) .
note that we will also quote a free - streaming mass , which is the mass at the mean density enclosed in a half - wavelength mode corresponding to @xmath21 . for the sake of simplicity ,
the analysis here is presented in terms of the mass of a thermal relic dark matter particle , for which there is a one - to - one correspondence between the free - streaming length and the particle mass .
we should point out , however , that in recent years sterile neutrinos and other non - thermal particles have become popular wdm candidates .
some of these models are actually more similar to mixed dark matter models with cold and warm dark matter components .
the shape of the free - streaming `` cut - off '' can then be quite different from that of a thermal relic , and may instead correspond to a downward step in the power spectrum rather than a cut - off ( see @xcite ) .
there is also no universal relation between free - streaming length and mass of the wdm particles in these models , and the normalization and functional form of this relation varies greatly between different non - thermal wdm candidates .
unfortunately , this has led to considerable confusion in the literature when wdm models , characterized by their model - dependent wdm particle masses , are compared between each other and/or thermal relic models and in particular with lyman-@xmath0 forest data .
for example ref .
@xcite quote a sterile neutrino mass of 2kev for their thermal relic wdm model which corresponds , however , to a thermal relic mass of 1.4 kev . for convenience and ease of comparison with the literature , in this work we therefore consider only matter power spectra with cut - off shapes expected for thermal relic wdm particles and quote the unambiguous thermal relic masses ( and corresponding cut - off scale ) to characterize our wdm models .
the lyman-@xmath0 forest , due to its spectral nature , probes the matter power spectrum in velocity space . with increasing redshift the ratio of a given ( comoving ) free - streaming length in velocity space to the thermal cut - off length scale at a given temperature increases as @xmath22 .
there is furthermore strong observational evidence that the temperature of the igm decreases toward higher redshift over the range @xmath23 @xcite , and therefore a corresponding decrease in the thermal cut - off length scale . despite observational difficulties , pushing to high
redshift allows the models to probe smaller free - streaming lengths and thus to improve the limits on wdm masses .
the high - redshift regime has also the advantage of probing structures that are more linear and the wdm cut - off is more prominent at high redshift compared to low redshift @xcite .
lyman-@xmath0 forest data to constrain wdm were first used in ref .
@xcite where a limit of 750 ev was obtained by using n - body simulations only . in previous work ,
@xcite , we used instead two samples of high - resolution qso lyman-@xmath0 forest spectra at @xmath24 to set a lower limit of 550 ev for the mass of a thermal wdm candidate . following this , ref .
@xcite and ref .
@xcite , using higher - redshift qso spectra from the sloan digital sky survey ( sdss ) and applying a different analysis method , significantly improved this limit by a factor @xmath25 . as already noted , however , care has to be taken in the correct modeling of the free - streaming properties of `` non - thermal '' candidate wdm particles of a given model - dependent mass , such as the popular sterile neutrino . in ref .
@xcite the authors have focused on constraints on a range of such models . because of a non - zero mixing angle between active and sterile flavour states , x - ray flux observations can also constrain the abundance and decay rate of such wdm particles ( e.g. @xcite ) .
the joint constraints from lyman-@xmath0 forest data and those from the x - ray fluxes of astrophysical objects now put considerable tension on the parameter space allowed for a sterile neutrino particle with the phase - space distribution proposed by dodelson & widrow @xcite , although other , possibly more phsyical scenarios should be explored @xcite . in ref .
@xcite we presented the most stringent lyman-@xmath0 forest limits up to that date on the free - streaming of dark matter , m@xmath26 4 kev ( 2@xmath27 ) .
that analysis was based on an ( at the time ) unrivaled sample of high - quality , high resolution qso absorption spectra extending to @xmath28 .
the limit is in obvious conflict with many of the recent suggestions for alleviating the difficulties encountered by numerical models in reproducing the observed properties of local group satellite galaxies within the cold dark matter paradigm .
these models often assume dark matter to be made up by thermal relic wdm with masses in the range 0.5 - 2 kev ( e.g. @xcite ) . since our study in @xcite , the size of our high - quality , high - redshift qso absorption spectra sample , the quality and size ( in particular the dynamic range and resolution ) of our numerical simulations and our knowledge of the thermal and ionization state of the igm at the relevant redshifts
have all significantly improved .
motivated by these improvements , and in light of the lively debate of dark matter possibly being warm with masses in the range 0.5 - 2 kev , we present here a new and much more extensive study of the high - redshift lyman-@xmath0 forest constraints on the free - streaming properties of dark matter .
the new study is based on an improved data set , further refined modeling of the flux power spectrum and a large suite of new numerical hydrodynamical simulations .
we also perform a comprehensive investigation of the the systematic uncertainties related to this measurement .
finally , it is worth highlighting that wdm would have profound implications in many astrophysical and cosmological contexts . in this respect ,
igm constraints are highly complementary to other probes based , for example , on the properties of dark matter haloes @xcite , the number of satellites and their luminosities @xcite , strong lensing , the velocity function in the local environment @xcite , phase - space density constraints @xcite , the formation of the first stars @xcite , the high - redshift quasar luminosity function @xcite , decays of wdm particles in the high redshift universe @xcite , reionization @xcite , gamma ray - bursts @xcite , galaxy formation aspects @xcite using n - body / hydrodynamical simulations @xcite or analytical / semi - analytical methods @xcite . the paper is organized as follows . in section [ data ]
we present our new data set .
the simulations are described in section [ sims ] .
the mock quasar sample , which will be important for estimating error amplitude and covariance , is introduced in section [ mock ] .
section [ fluxpower ] discusses the effect of the most important physical parameters on the flux power spectrum , while most of the remaining nuisance parameters and the impact they have in terms of flux power are discussed in an appendix .
our main results are reported in section [ method ] , together with a description of the monte carlo sampling of the likelihood space .
we summarize our findings and conclude in section [ conclu ] .
our analysis is based on high - resolution spectra of 25 quasars with emission redshifts @xmath29 . compared to our previous analysis in ref .
@xcite the number of qso spectra , at these redshifts , has improved by nearly a factor two .
spectra for fourteen of the objects were taken with the keck high resolution echelle spectrometer ( hires ) @xcite , and the remaining eleven were taken with the magellan inamori kyocera echelle ( mike ) spectrograph on the magellan clay telescope @xcite . most of the data have been presented elsewhere @xcite . here
we briefly review the relevant features of the spectra , and describe how the flux power spectra were calculated .
the majority of spectra were reduced using a custom set of idl routines based on optimal sky subtraction @xcite and optimal extraction @xcite techniques , while a small subset of the hires spectra ( pss 0248@xmath301802 and br 1202@xmath310725 ) were reduced using the makee software package . the hires and mike spectra have spectral resolutions of 6.7 and 13.6 kms@xmath32 ( fwhm ) , and the spectra were extracted using 2.1 and 5.0 kms@xmath32 spectral bins , respectively .
the one - dimensional relative flux - calibrated spectra were then continuum normalized using spline fits based on power - law extrapolations of the continuum redward of the lyman-@xmath0 emission line .
median continuum signal - to - noise ratios within the lyman-@xmath0 forest of each object are typically in the range of @xmath33 per pixel .
the continuum estimates are necessarily crude due to the high levels of absorption in the lyman-@xmath0 forest at these redshifts .
we estimate that typical uncertainties in the continuum are of the oder @xmath1410 - 20% , a point we return to in the power spectrum analysis
. to compute the flux power spectra , we first divided the lyman-@xmath0 forest in each quasar spectrum into two regions of equal redshift length .
we then computed the power spectrum of the fractional transmission , @xmath34 , in each region separately , where @xmath35 here , @xmath36 is the mean transmitted flux calculated at the mean redshift of each region .
we used fixed relations for the mean flux given by @xmath37}$ ] , where @xmath38 the fit to @xmath39 at @xmath40 is from @xcite , while the evolution at @xmath41 is based on a fit to the mean flux measured from the data presented here .
the latter is similar to the trend in @xmath42 presented by @xcite .
we note that our analysis is not sensitive to our choice of using a fixed relation for @xmath36 . in tests where we instead divided each region by the mean flux in that region alone we obtained very similar power spectrum estimates on average . when calculating the flux power spectrum we do not attempt to mask metal lines ( see discussion below ) .
we do , however , mask regions of strong telluric absorption ( @xmath43 , @xmath44 , @xmath45 , @xmath46 , and @xmath47 ) .
the longest - wavelength mask effectively means that we probe up to a maximum redshift of 5.684 .
the power spectra for individual regions were averaged over ten logarithmic wavenumber bins in the range @xmath48 $ ] with 0.2 dex spacing .
the power spectra from all regions were then further averaged according to instrument and the mean redshift in each region .
we used median redshifts @xmath49 for a nominal total of eight combined power spectra and 80 data points . in order to be conservative
, however , we decided to use a subset of this sample and do not consider the highest redshfit bin for the mike data set ( which has very large error bars ) or the flux power measurements at @xmath50 , which might be affected by continuum fitting uncertainties .
the final data set used in the present analysis thus consists of 49 data points .
preliminary estimates of the error in the power spectra were calculated using a bootstrap approach .
it is known , however , that bootstrapping typical underestimates the true errors ( e.g. @xcite ) . to be conservative , we therefore decided to add an additional 30% uncertainty to our estimates of the errors of the observed flux power spectrum for our standard analysis .
we will also quote the tighter limits that would be obtained without this increase of the error estimate . as a further check
, we used the set of mock qso spectra described in section [ mock ] to determine what the expected covariance in the power spectra should be ( within the limits of our finite simulation box ) at each redshift for a sample of similar size and quality to the one used here .
these estimate were used to correct a few error estimates in the real data that appeared to be too small . with these corrections ,
the final flux power spectra used here have error bars that are larger than @xmath51 .
we model the flux power spectrum based on a set of hydrodynamical simulations performed with a modification of the publicly available code .
this code implements a simplified star formation criterion @xcite that turns all gas particles that have an overdensity above 1000 and a temperature below @xmath52 k into star particles .
this has been first used and extensively tested in ref .
@xcite .
the reference model , hereafter referred to as @xmath53 , is a box of length 20 @xmath54 comoving mpc with @xmath55 gas and cold dm particles ( with a gravitational softening length of 1.3 @xmath54 kpc ) in a flat @xmath8cdm universe with cosmological parameters @xmath56 , @xmath57 , @xmath58 and @xmath59 , in agreement both with wmap-9yr and planck data @xcite .
we further explore three different wdm models with masses @xmath60 kev ; these models correspond to 50% suppression of power in the linear ( redshift independent ) matter power spectrum at scales @xmath61/mpc , respectively .
the initial condition power spectra are generated with camb@xcite and the suppression and velocity for the wdm particles are implemented using the approach outlined in ref .
in order to assess convergence and evaluate resolution corrections ( which are model dependent ) , we also perform four additional @xmath62 models , one each for the reference and wdm cases , and a single @xmath63 model for the 2 kev simulation only .
we also have performed three @xmath64 simulations for @xmath8cdm , wdm 1 kev and wdm 2 kev in order to check the flux power spectrum convergence at the largest scales of our smaller boxes .
we explore the impact of different thermal histories on the lyman-@xmath0 forest by modifying the ultra violet ( uv ) background photo - heating rate in the simulations ( e.g. @xcite ) .
a power - law temperature - density relation , @xmath65 , arises in the low density igm ( @xmath66 ) as a natural consequence of the interplay between photo - heating and adiabatic cooling @xcite .
we consider a range of values for the temperature at mean density , @xmath67 , and the power - law index of the temperature - density relation , @xmath68 , based on the observational measurements presented recently by ref .
these consist of a set of 3 different indices for the temperature - density relation , @xmath69 , that are kept approximately constant over the redshift range @xmath70 $ ] and 3 different temperatures at mean density , @xmath71k , which evolve with redshift , yielding a total of 9 different thermal histories .
the reference thermal history assumes @xmath72 .
these 9 thermal histories have been performed for all the 3 wdm models and for the reference @xmath8cdm case , resulting in a total of 36 simulations .
in addition to these parameters we also consider and vary several other physical parameters for the reference model only , given that these are poorly constrained by the data .
these are the redshift of reionization @xmath73 ( i.e. the redshift at which the optically thin uv background is switched on in the simulations ) which is chosen to be @xmath74 for the reference case and @xmath75 for two additional models ; the hubble constant , with two extra simulations with @xmath76 ; the scalar spectral index , with @xmath77 ; the matter content , with @xmath78 and the r.m.s
. amplitude of the matter power spectrum , with @xmath79 .
we note here that varying the redshift of reionization in particular enables us to assess the impact of different integrated thermal histories ( i.e. the effect of jeans smoothing , see @xcite for a recent discussion ) on our analysis .
the effect of different integrated thermal histories on lyman-@xmath0 forest constraints was also considered in ref .
@xcite using this parameterisation .
overall , a total of 54 hydrodynamical simulations have been performed .
approximately 4000 core hours were required for each ( 20,512 ) run to reach @xmath80 , with the higher resolution simulations requiring around 5 times longer . during
the simulation runs , we extract the non - linear matter power spectra in order to compare with @xcite . for the reference case
only , we additioanlly extract the position of the haloes with a friends - of - friends halo finding algorithm for our model of the impact of spatial fluctuations in the uv background on the flux power ( see the appendix for further details ) . ratio between the 3d non - linear matter power spectrum of 3 different wdm models ( 1 , 2 and 4 kev , black , blue and orange curves ) at 3 different redshifts ( @xmath81 , represented by the dot - dashed , dashed and continuous curves ) and the corresponding @xmath8cdm model .
the green curve represents the linear redshift independent suppression in terms of matter power for a @xmath82 kev model obtained using eq .
6 of ref .
the arrows in the bottom part of the figure indicate the maximum value of the wavenumbers probed by the sdss data and by the data set used in the present analysis .
this figure refers to the reference ( 20,512 ) simulations . ]
[ fig_b ]
lastly , we note that the physical properties of the lyman-@xmath0 forest obtained from the treepm / sph code gadget - ii are in very good agreement at the percent level with those inferred from the moving - mesh code arepo @xcite and with the eulerian code enzo @xcite .
the simulated lyman-@xmath0 forest spectra are extracted along 5000 random line - of - sights ( loss ) after interpolation of the relevant physical quantities along the loss using the sph formalism .
box - size effects on the flux power are estimated with @xmath64 simulations .
note , however , that the necesary box size correction is below the percent level at the largest scales used .
resolution corrections are , however , important .
the flux power spectra are corrected for resolution effects using the @xmath62 simulations ( see the appendix for further details ) .
the mean flux is varied _ a posteriori _ , after having extracted the spectra , by reproducing @xmath83 times the observed @xmath84 ( see appendix ) . at the end of the procedure the four - dimensional parameter space in @xmath85 is explored fully by means of quadrilinear interpolation performed over the set of 36 hydrodynamical simulations and 108 ( @xmath86 mean flux values ) flux models . in order to get a better understanding of the expected ( co)variance properties of the observed data we generate samples of mock qso absorption spectra which resemble the observational data as closely as possible .
the procedure used to create the mock spectra can be summarized as follows : @xmath87 we consider the total redshift path in each redshift bin and combine the short simulated spectra ( @xmath88 mpc@xmath89 in length ) to match the total length of an observed qso spectrum ( approximately 40 spectra are used ) ; @xmath90 we allow for an optical depth evolution along the los ( which is absent since our simulated spectra are from snapshots at fixed redshifts ) following the scaling expected from the fluctuating gunn - peterson approximation , @xmath91 ( see e.g. @xcite ) ; @xmath92 for each short simulated spectrum we consider a @xmath93 20% error on the quasar continuum placement ( the continuum is drawn randomly from a gaussian distribution around the value 1 with a @xmath94 ) ; @xmath95 we smooth the flux with a gaussian at a given fwhm corresponding to the spectrograph resolution and rebin the spectra with the observed pixel - size ; @xmath96 we add gaussian - distributed noise on top of the flux , matching the signal - to - noise of the observational data .
we demonstrate in the appendix that the ( instrumental ) effects of noise and finite resolution , which are scale and redshift dependent , are below 20% ( 6% ) at the smallest scales for mike ( hires ) .
the ( co)variance properties of this mock sample are in reasonable agreement with those of our observed sample , both as a function of redshift and wavenumber .
there are only 5 data points that appear to have error bars that are smaller than those obtained from the mock sample : 4 data points from the mike sample ( @xmath97(s / km)@xmath98 at @xmath99 , @xmath97(s / km)@xmath100 at @xmath101 , @xmath97(s / km)@xmath102 at @xmath103 ) and one data point from the hires sample ( @xmath104 at @xmath99 ) . as the observed sample is still small and
it is thus expected that the bootstrap errors estimated from the data could be unrealistically small , for these data points we increase the error bars to match those obtained from a mock sample of 30 qsos in the same redshift bin .
these mock data points , with the extra 30% error added , otherwise agree well with those of the observational data , giving us confidence that this is reasonable .
finally , we note that the hydrodynamical simulations used to construct our mock lyman-@xmath0 forest spectra do not incorporate chemical elements other than hydrogen and helium .
we have therefore estimated how unidentified , lower redshift metal lines in the lyman-@xmath0 forest may bias our result , and in particular how these narrow absorption lines may alter the flux power spectrum at small scales .
furthermore , the simulations also assume a spatially uniform uv background .
we therefore also estimate the impact of spatial fluctuations in the uv background on the lyman-@xmath0 forest at @xmath99@xmath105 . we follow a modified version of the approach described in ref .
@xcite for exploring fluctuations in the ionising background at lower redshift , to which we refer the reader for further details .
these two systematics effects will also be discussed more extensively in the appendix .
in this section we demonstrate the distinctively different effects that the thermal ( i.e. due to the temperature of the photo - ionized igm ) and wdm cut - off have on the flux power spectrum .
we will also check for possible effects due to the limited numerical resolution of our simulations .
firstly , however , it is instructive to consider fig .
[ fig_a ] , where we show the ratio of the non - linear matter power spectra in the wdm and @xmath8cdm simulations .
the results are shown at three different redshifts , @xmath106 .
the redshift range @xmath99@xmath107 brackets that of the high - resolution data set used in this work .
we additionally present the matter power spectrum at @xmath108 to show the evolution of the non - linear power at lower redshift .
the three wdm models are reported as orange , blue and black curves for masses of 4,2 and 1 kev , respectively , while the green curve shows the _ linear _ suppression for the 2 kev case taken from ref .
the power spectra are already clearly somewhat non - linear at high redshift ; the blue and green curves start to differ significantly at small scales at @xmath109mpc . in the bottom part of the panel
we show the approximate wavenumber ranges that are probed by sdss and the hires+mike data set used in our analysis .
note that the non - linear matter suppression is in good agreement with the fitting formula presented in ref .
@xcite . in figure [ fig_b ]
we qualitatively compare a set of noiseless lyman-@xmath0 forest spectra extracted from the @xmath8cdm , wdm 1 kev and wdm 2 kev models , represented by the green , black and blue curves respectively .
it is clear that the amount of small - scale substructure in the transmitted flux in the @xmath8cdm is more prominent with respect to the wdm cases . in the rest of this section
we will quantify these differences in terms of the 1d flux power spectrum .
we now turn to figure [ fig_c ] , which shows the ratio between the 1d flux power of the wdm and @xmath8cdm models for the four different redshift bins used in the present analysis ( note that we compute the power spectrum of the quantity @xmath110 , and we refer to this as the flux power ) .
the suppression of the flux power is larger than that seen in the matter power spectrum .
this is due to the fact that the 1d matter power spectrum is an integral of the 3d power spectrum and therefore very sensitive to the small scale cut - off .
as expected , the largest differences exist between the 1 kev ( black curves ) and the @xmath8cdm model .
note that the flux power also changes at large scales ; the requirement of reproducing the same observed mean flux value ( given by eq .
4 ) results in an increase of the power at those scales ( the power spectrum of the wdm _ flux _ @xmath111
, not @xmath112 , does show suppression over all scales when compared to @xmath8cdm ) .
furthermore , we also note that there is a substantial redshift evolution of the flux power between @xmath113 and @xmath99 . the ratio of the 1d flux power spectrum for 3 different wdm models ( 1 , 2 and 4 kev represented in black , blue and orange ) at 4 different redshifts ( @xmath49 represented by the continuous , dotted , dashed and dot - dashed curves , respectively ) to the corresponding @xmath8cdm flux power spectra .
this figure displays the results from the ( 20,512 ) simulations .
the mean flux is the same in all models and the shaded area shows the range of wavenumbers used in the present analysis . ]
the ratio of the 1d flux power spectrum for 2 different wdm models ( 1 and 2 kev , represented in black and blue ) at 2 different redshifts ( @xmath114 represented by the continuous and dot - dashed curves , respectively ) to the corresponding @xmath8cdm simulations .
the thin curves refer to the ( 20,512 ) simulations , while the thick curves refer to the high resolution ( 20,768 ) models .
the mean flux is the same for all models and the shaded area shows the range of wavenumbers used in the present analysis . ]
numerical convergence for wdm simulations can be particularly difficult to achieve ( see ref .
@xcite ) . in figure [ fig_d ]
we demonstrate that at the resolution and wdm masses considered in this work , this should , however , not be an issue .
figure [ fig_d ] compares the flux power extracted from the @xmath53 and @xmath62 1 and 2 kev simulations to the corresponding @xmath8cdm simulations at the same resolution .
the agreement between the different resolution simulations is very good , typically at the percent level .
the differences are largest for the 1 kev case at the smallest scales probed by our data in the current analysis , where they reach the 10% level . the simulated flux power spectra for both @xmath8cdm and wdm models
have therefore been corrected for resolution effects by multiplying the raw power spectra by the ratio of the results from the @xmath62 and @xmath53 simulations . in general
, we find the requirements for reaching numerical convergence in terms of flux power are more demanding for absolute values of the flux power rather than ratios of different models w.r.t . the @xmath8cdm case .
this will be discussed further in the appendix . in figures
[ fig_f ] and [ fig_g ] we explore the effects of the two thermal parameters , @xmath115 and @xmath68 , on the flux power spectrum .
as discussed earlier , the @xmath116 relation is usually parameterized as a power - law , @xmath117 . in both of these figures
we also plot the wdm 2kev model in order to emphasize the very distinct differences between the thermal and wdm cut - offs , both in the dependence on wavenumber and redshift .
a hotter ( colder ) @xmath115 value produces a suppression ( enhancement ) in the flux power spectrum with a redshift dependent cut - off .
the wdm cut - off is instead more pronounced and steeper than the cut - off induced by a hotter igm .
the dependence of the thermal cut - off on the slope of the temperature - density @xmath68 is fig .
[ fig_g ] is also very different from the wavenumber and redshift dependence of the wdm cut - offs , and is much flatter over the wavenumber range considered here .
the ratio of the 1d flux power spectrum for two @xmath8cdm models with different temperatures(hot , roughly hotter by 3000 k with respect to the reference simulation , in orange and cold , roughly colder by 3000 k with respect to the reference simulation , in black ) and at four different redshifts ( @xmath49 represented by the dot - dashed , dashed , dotted and continuous curves , respectively ) to the corresponding @xmath8cdm simulations .
the wdm 2 kev model is also shown in blue .
the mean flux is the same for all models , and the shaded area shows the range of wavenumbers used in the present analysis . ]
the ratio of the 1d flux power spectrum for two @xmath8cdm models with different slopes of the temperature - density relation ( @xmath118 in orange and @xmath119 in black ) and at four different redshifts ( @xmath49 , represented by the dot - dashed , dashed , dotted and continuous curves , respectively ) to the corresponding @xmath8cdm simulations .
the wdm 2 kev model is also shown in blue .
the mean flux is the same for all models , and the shaded area shows the range of wavenumbers used in the present analysis . ]
the effects due to to changing the mean flux level are discussed in detail in the appendix ( fig .
[ fig_gg ] ) .
we conservatively assume a range of @xmath120 for the observed effective optical depth .
we further note that the dependence of changing the mean flux level on wavenumber is even flatter than that obtained for variations of @xmath68 , but shows a weak scale dependence in the highest redshift bin . in this section
we now briefly discuss the following systematic effects : instrumental resolution ; noise ; spatial fluctations in the uv background and metal line contamination . in the appendix
there is a more detailed description of these nuisance effects and how they are modeled .
we only summarize the main quantitative results here .
instrumental resolution , which is different for the two sub - data sets , suppresses the flux power spectrum by at most 20% and 5% for mike and hires , respectively , at the smallest scales probed , with a negligible redshift dependence .
the signal - to - noise ratio impacts at about the 2 - 3% level at the smallest scales for @xmath121 while it is at the 7% level for the highest redshift bin .
the uv background fluctuations have been implemented with a deliberately extreme model based on ionizing emission from quasars only .
the impact of this extreme model of uv fluctuations on the flux power spectrum is quite scale dependent , and rises considerably at large scales ( see also @xcite ) . at the scales of interest here
the effect on the flux power spectrum is below the 10% level ( see figure [ fig_h ] ) .
this should be considered as a generous upper limit .
the metal contamination has a much smaller effect on the flux power spectrum , below the 1% level for the whole range of scales considered ( see figure [ fig_hh ] ) .
apart from the apparently negligible metal contamination , the other nuisance effects have been fully implemented in our analysis . in table
i , we summarize the main nuisance parameters and present rough estimates of the relative errors induced in the flux power spectrum . where possible , effects with a known amplitude such as resolution and noise characteristics are simply incorporated into the mock qso spectra .
the remaining parameters are fully marginalized over in our likelihood analysis .
@xmath122 & notes + data res . &
@xmath123% & corrected + data s / n & @xmath124% & corrected + num .
& @xmath125% & corrected + metals & @xmath126% & neglected + mean flux & @xmath127% & marginalized + thermal history & @xmath127% & marginalized + uv & @xmath128% & marginalized +
we now turn to our analysis of the data and discuss our results . for all 108 flux models considered in our analysis , we compute the ratio of these models with respect to the reference model .
we then bin this ratio at the same wavenumbers as the data . in practice , this means we have a 4d parameter matrix with @xmath129 that summarizes all the results obtained from the hydrodynamical simulations , plus some further parameters for which we have established the effect on the flux power for the reference model only .
we parameterize the effect of uv background fluctuations on the flux power with a factor @xmath130 that multiplies the flux power spectrum corrections shown in fig .
[ fig_hh ] , constrained to be in the range @xmath131 $ ] and applied in addition to the corrections discussed in the previous section ( @xmath132 means that the power spectrum is corrected exactly by the amount shown in the lower panel of fig .
[ fig_hh ] ) .
we decide to neglect the effect of metal contamination since it is , as we discuss further in the appendix , very small .
we then perform second order taylor interpolations for the following remaining parameters : @xmath133 , as in refs .
@xcite . in figure [ fig_i ]
we compare our best - guess model ( the reference simulation ) , represented by the orange curves , with the observational data .
this best - guess model will be the reference point of our likelihood code that will be described below .
note that there is already rough `` visual '' agreement with the data , albeit with a poor @xmath134 value .
the two - dimensional 1 and 2@xmath27 contours for mean ( in colour ) and marginalized ( solid black curves ) likelihoods for the parameters @xmath135 against @xmath136 and @xmath135 against @xmath137 obtained from the mike+hires data sets .
these results assume a power - law evolution for @xmath115 and @xmath68 , but with @xmath138 constrained to be in the [ 0.7,1.7 ] range , and refer to a run for which some planck - like priors on @xmath137 , @xmath139 and @xmath140 have been applied .
note , however , that our results are not sensitive to this choice of prior . ] the one - dimensional mean ( dotted curve ) and marginalized ( continuous curve ) likelihoods for the parameter 1 kev/@xmath141 .
these results refer to a run for which some planck - like priors on @xmath137 , @xmath139 and @xmath140 have been applied .
note , however , that our results are not sensitive to this . ]
we use a modified version of the code @xcite to derive parameter likelihoods from the lyman-@xmath0 forest data . for the hires+mike data
, we have a set of 15 parameters : 6 cosmological parameters ( @xmath142 , @xmath143 , @xmath141 ) ; 4 parameters describing the thermal state of the igm , using a power - law parameterization of the temperature - density relation , @xmath144 , with parameters @xmath145^{t_0^s}$ ] and @xmath146^{\gamma_a^s}$ ] ; 4 parameters describing the evolution of the effective optical depth with redshift , since a single power - law has been shown to be a poor approximation over this wide redshift range ( see @xcite ) and one parameter describing the spatial fluctuations in the uv background @xmath147 .
we apply strong gaussian priors to @xmath148 in order to mimic planck constraints : @xmath149 .
we have checked that these priors do not affect any of our constraints on the free - streaming length / wdm mass , but they are helpful in obtaining faster convergence of the monte - carlo chains .
we vary @xmath150 in the range @xmath151 $ ] k and @xmath152 in the range @xmath153 $ ] , and thereby heavily penalize the @xmath134 if @xmath154 is outside the physical range @xmath153 $ ] .
the values of @xmath155 and @xmath143 are not constrained by the data and they are prior dependent : the range chosen are @xmath156 $ ] km / s / mpc and @xmath157 $ ] , respectively .
the covariance matrix calculated from our data set is noisy ( especially at high redshift ) , preventing a reliable inversion .
we have therefore regularized the observed covariance matrix with the correlation coefficients as estimated from the simulated spectra as in ref .
@xcite , @xmath158 with @xmath159 , where @xmath160 and @xmath161 are the covariance matrices of the observed and simulated spectra , respectively .
lccc parameter & ( 1@xmath27 ) & ( 2@xmath27 ) & best fit + @xmath139 & [ @xmath162 & [ @xmath163 & 0.957 + @xmath137 & [ @xmath164 & [ @xmath165 & 0.822 + @xmath140 & [ @xmath166 & [ @xmath167 & 0.298 + @xmath168 & [ @xmath169 & [ @xmath170 & 1.16 + @xmath171 & [ @xmath172 & [ @xmath173 & 1.32 + @xmath174 & [ @xmath175 & [ @xmath176 & 1.91 + @xmath177 & [ @xmath178 & [ @xmath179 & 3.09 + @xmath180 & [ @xmath181 & [ @xmath182 & 1.64 + @xmath183 & [ @xmath184 & [ @xmath185 & -0.15 + @xmath186 k & [ @xmath187 & [ @xmath188 & 9.2 + @xmath189 k & [ @xmath190 & [ @xmath191 & -2.5 + @xmath130 & [ @xmath192 & [ @xmath192 & 0.18 + @xmath143 & [ @xmath193 & [ @xmath194 & 11.2 + @xmath195 kev/@xmath196 & [ @xmath197 & [ @xmath198 & 0.03 + our results are summarized in table [ tab2 ] .
we obtain a @xmath3 upper limit on the parameter 1kev@xmath199 of 0.3,which translates into the following constraints : @xmath200 kev at the @xmath3 c.l . and @xmath201 kev at the @xmath202 c.l .
, with a best - fit value of @xmath203 kev .
for a @xmath204 kev wdm particle the 50% suppression in the 3d linear matter power compared to the @xmath8cdm case matter power spectrum is at a scale of @xmath205mpc , while the suppression at @xmath206mpc is about 10% .
if we drop the 30% additional error applied to the observed flux power spectrum ( see section ii ) we get a tighter lower limit of @xmath207 kev ( @xmath3 ) .
the @xmath134 gets worse by @xmath208 , but still has a probability of 11% of being this large for the present number of degrees of freedom .
we also took a `` frequentist '' approach and fixed the values of @xmath141 to 2.5 and 3.3 kev and found the results in terms of the other parameters : in this case the @xmath134 is of course higher than in the @xmath8cdm model ( with @xmath209 , respectively ) but nevertheless compatible with the results obtained in our standard analysis .
this is similar to the approach used in ref .
@xcite , in which an analysis of mixed cold and warm models was performed in both a bayesian and in a frequentist approach . the degeneracies between the parameter 1kev@xmath199 and the other parameters are very weak .
[ fig2d_b ] we show the 2d contour plots for the mean likelihood ( in colour ) and the marginalized likelihood ( black curves ) for @xmath210 and @xmath137 versus 1 kev/@xmath211 . in fig .
[ fig1d ] we report the 1d mean and marginalized likelihoods for 1kev@xmath199 ( continuous and dotted curves , respectively ) .
the redshift evolution of the temperature at mean density , @xmath115 , used in our reference model is shown as continuous black curve , while the the two dashed line display our cold and hot models .
recent measurements of the igm temperature at mean density obtained by ref .
@xcite are also shown for different values of @xmath68 .
the measurement at @xmath212 is taken from @xcite .
the results of our likelihood analysis are shown with the shaded orange area ( for the power - law evolution case , @xmath213 ranges ) , and with orange triangles for model where we left the temperature free in the four redshift bins ( @xmath202 error bars ) . in both cases the temperature values reported are the marginalized results .
these results refer to a run for which some planck - like priors on @xmath137 , @xmath139 and @xmath140 have been applied . ]
we obtain the following evolution for the temperature - density relation @xmath214^{-2.5}$ ] k and @xmath215^{-0.15}$ ] .
the inferred temperature is decreasing with increasing redshift , while the redshift evolution of @xmath68 is weak .
we stress that the igm thermal state is just one of several nuisance parameter in our likelihood analysis over which we marginalise .
we discuss it here in the context of a consistency check rather than as a measurement . with this in mind , in figure [ fig_therm ]
we show the recovered redshift evolution for @xmath115 compared to three input thermal histories used in the simulations and measurements obtained from high resolution lyman-@xmath0 forest data from refs .
@xcite ( note that the power - law index of the temperature - density relation , @xmath68 , has not yet been measured directly at @xmath216 ) .
the shaded orange area brackets the @xmath213 of the temperatures obtained by our standard likelihood analysis after marginalization .
the inferred temperature evolution ( parameterized as a power law in redshift ) is in good agreement with the measurements from ref .
we also tested a model where the igm temperature is left to vary freely in the four redshift bins , shown by the orange data points with error bars in figure [ fig_therm ] .
in the two highest redshift bins this analysis returns temperatures that are rather cold and are disfavored by the data with an unreasonably large temperature jump between @xmath103 and @xmath217 ( see fig . [ fig_therm ] ) .
this suggests that in this case we have introduced too many free parameters and are most likely `` overfitting '' the flux power spectrum .
for completeness we mention that with this apparently unphysical temperature evolution our analysis gives a constraint on @xmath141 ( @xmath3 c.l . ) which is about 1 kev lower than for our standard analysis and also returns an unreasonably low reduced @xmath134 value .
finally we have also performed a likelihood analysis where the igm temperature is fixed to be unrealistically cold throughout ( 3000k , independent of redshift ) to allow for a maximum contribution of the free - straming of wdm to the observed cut - off in the flux power . again
just for completeness , for this model the constraint on @xmath141 ( @xmath3 c.l . ) is lower by about 0.5 kev compared to our standard analysis .
the recovered effective optical depth values at each redshift bin are usually within 20% of the measured optical depth evolution used as the input into the likelihood calculation . the inferred values for the amplitude and slope of the matter power spectrum and for the matter content do not show biases with respect to the planck - like priors we used .
overall the @xmath134 for the best fit model is 34 for 37 d.o.f . which has a reasonably high probability of about 60% of being larger than this value .
lastly , in figures [ figbf1 ] and [ figbf2 ] we show our final best - fit model compared to the data obtained with mike and hires , respectively .
the best fit model is shown as the green curves .
we also overplot , for comparison purposes only , three other models that are excluded with very high significance by the present analysis : a model which has a wdm mass of @xmath218 kev ( red curves ) and a hot model with a temperature value which has been increased by 3000 k with respect to the best fit case ( orange curves ) . when calculating the predicted flux power spectrum for these three models we change only one parameter each time and leave all other parameters fixed at their overall best fit values .
in addition , we show the case in which fix @xmath218 kev and allow all other parameters to assume their best - fitting values under this assumption ( blue curves ) . compared to our previous findings obtained in ref .
@xcite , it is worth stressing the main differences .
first of all , from the data side , the sample used here extends to high redshift and double the amount of spectra contributing to the signal at @xmath219 . secondly ,
both the simulations and the analysis have been refined by : increasing the number of hydrodynamical simulations and their resolution ; improving the method in a way that allows a full sampling of the most relevant parameter space ( thermal parameters , wdm cutoff and mean flux ) compared to a poorer sampling of the parameter space made in ref .
when considering only the high - resolution data set , we improve the limits by nearly a factor three from 1.2 kev to 3.3 kev at the @xmath3 c.l . , this is due to both the data and the modelling of the flux power .
in this section we present the joint analysis with the sloan digital sky survey ( sdss ) 1d flux power spectrum data of ref .
@xcite where the authors have presented the flux power spectrum of a sample of 3035 qso absorption in the redshift range @xmath220 drawn from the dr1 and dr2 data releases of sdss .
these data have a spectral resolution of r @xmath221 , and so typical lyman-@xmath0 absorption features , which have a velocity width of @xmath127 kms@xmath32 , are not resolved .
the wide redshift range , however , makes this data set very constraining in terms of cosmological parameters . as a final result of their analysis they present an estimate of flux power spectrum @xmath222 at 12 wavenumbers in the range @xmath223 ( s / km)@xmath224 , equally spaced in @xmath225 for @xmath226 for a total of 132 data points .
this measurement is likely to improve soon with the new analysis made by the sdss - iii team of a sample which is about 50 times larger than the one we utilize here @xcite . for the joint sdss+mike+hires analysis we have used a total of 28 parameters :
15 parameters as used for the hires / mike spectra ( without @xmath227 and the two parameters describing the effective optical depth evolution at @xmath103 ) plus 13 noise - related parameters : 1 parameter which accounts for the contribution of damped - lyman-@xmath0 systems and 12 parameters modeling the resolution and the noise properties of the sdss data set ( see @xcite ) .
we do not consider the possible effect of different reionisation scenarios on the sdss flux power .
the covariance matrix of the sdss flux power is provided by the authors of @xcite .
the @xmath3 lower limit on @xmath141 is unchanged at 3.3 kev but now with a @xmath228 for the best fit model for a total of 181 data points and 170 degrees of freedom , which has a 23% probability of being this large .
given the fact that the sdss and mike+hires data sets do not have redshift overlap there is no significant bias in the other recovered marginalized parameters .
note that for the joint analysis we used the sdss likelihood based on second order taylor expansion of the flux power as described in ref .
@xcite and used in ref .
@xcite . unlike our previous findings obtained in ref .
@xcite , where the wide redshift range of sdss data was helpful in breaking the degeneracies between thermal parameters and wdm cutoff , we notice that in this case the sdss data do not improve the overall constraints .
this means that the constraining power of the new high - resolution data set is higher than the low - resolution sdss data .
the joint analysis gives now a lower limit of 3.3 kev , compared to the previous 4 kev value is thereby slightly less stringent : this is due to the different interplay between the data sets and to the relative role of the degeneracies present between igm thermal state and wdm cut - off .
it is also important to stress that the 30% extra error budget on the high - resolution data impacts on the final results also for the joint analysis making the results less stringent than in ref .
@xcite , where this error was not present .
ccccc model & ( 1@xmath27 ) & ( 2@xmath27 ) & best fit & @xmath134/d.o.f .
@xmath229 kev & @xmath230 kev & 33 kev & 34/37 + ref .
w / o 30% & @xmath231 kev & @xmath232 kev & 100 kev & 48/37 + ref .
w / o covmat & @xmath233 kev & @xmath234 kev & 14.3 kev & 33.2/37 + ref .
+ sdss & @xmath235 kev & @xmath236 kev & 42 kev & 183.3/170 + we have presented a comprehensive analysis of the transmitted lyman-@xmath0 flux power spectrum extracted from a set of 25 high - resolution qso spectra taken with the hires and mike spectrographs .
this represents an improved and extended version of the sample originally analysed in @xcite .
the lyman-@xmath0 forest is an excellent probe of the matter distribution at intermediate and high redshift in the mildly non - linear regime , from sub - mpc up to bao scales . in this work
we have focused on constraining any possible suppression of the total matter power spectrum which could be induced by the free - streaming of wdm particles in the form of a thermal relic . due to the non - linear nature of the the relationship between the observed lyman-@xmath0 flux and underlying matter density ,
departures from the standard @xmath8cdm case are expected over a range of scales that span at least one decade in wavenumber space and can be constrained by the data used in the present analysis .
we model this suppression by using a set of high - resolution hydrodynamical simulations and by marginalizing over a large range of physically motivated thermal histories
. the wdm cut - off exhibits a distinctive behavior which we demonstrate is not degenerate with other physical effects due to its different redshift and scale dependence .
we consider possible sources of systematic errors including metal line contamination , spatial fluctuations in the uv background intensity and uncertainties in the mean flux level estimation .
galactic feedback either in the form of supernova driven galactic winds or active galactic nuclei ( agn ) feedback should not impact the flux power spectra at the high redshift considered in this analysis @xcite .
our final results are obtained by means of a monte carlo markov chain likelihood analysis around a best - guess reference model .
the constraints quoted for @xmath141 have been calculated after marginalization over the other astrophysical and cosmological parameters .
our analysis is conservative in the following sense : we have dropped the estimates of the power spectrum at the largest scales probed by our sample in order not to be sensitive to continuum fitting uncertainties ; we add an additional error of about 30% to our error estimates obtained by bootstrapping to account for the expected underestimation of the real error ; and we allow for large fluctuations in the uv background fluctuations , which appear to be the most important nuisance factor .
furthermore , we create a mock qso sample which resembles as closely as possible the real data including noise and resolution and use the covariance matrix of this mock sample as an estimate of the error properties of the real data .
our final result of this analysis is @xmath237 kev at the @xmath3 c.l . , where the mass refers to that of a thermal relic .
this mass implies that wdm models for which there is a suppression in terms of the 3d linear matter power at scales @xmath238/mpc ( @xmath239/mpc ) larger than 10% ( 50% ) when compared to the @xmath8cdm case , are disfavoured by the present data sets .
the corresponding value of the `` free - streaming '' mass is @xmath240 . a model with a 2.5 kev thermal relic mass
is disfavoured by the data at about @xmath6 c.l . , a 2 kev mass at about @xmath5 c.l . , and a @xmath241 kev model at about @xmath4 c.l .
our final marginalized estimates and best fit values for @xmath141 are summarized in table iii .
overall , the final results presented are similar to those we have obtained in our previous analysis ref .
@xcite [ 3.3 ( 4.5 ) kev vs 4 kev previously if we include ( do not include ) an additional 30% error to account for a possible underestimate of the statistical error from a boot - strapping analysis ) .
we emphasize , however , that the present analysis is considerably more robust .
it uses a larger data set , a much improved analysis based on a broader suite of significantly improved simulations and as well as an extensive analysis of the systematic uncertainties .
further improvement of the constraints on the free - streaming of dark matter particles from lyman-@xmath0 forest data could come mainly from an enlarged set of high - quality , high resolution spectra , especially at the highest redshifts where the flux power spectrum is most sensitive to the free - streaming of dark matter .
an increase of the dynamical range of the simulations and improved independent constraints on the thermal state and thermal history of the igm are next on the list as requirements for further corroborating and perhaps pushing the constraints to even larger thermal relic masses . in the future ,
considerably stronger constraints on wdm may be derived using a baryonic tracer which is colder than the photo - ionized igm , thus moving the thermal cutoff to smaller scales in the flux power spectrum .
studies of 21 cm absorption / emission by neutral hydrogen gas before reionization , for example , could eventually fulfill this requirement .
however , with a lower limit of @xmath242m@xmath243 for the mass of dark matter haloes whose abundance could still be significantly affected , the lyman-@xmath0 forest data appears to leave already very little room for a contribution of the free - streaming of warm dark matter to the solution of what has been termed the small scale crisis of cold dark matter . in particular
, recent suggestions for models with relic masses of 0.5 - 2kev are significantly disfavoured by our analysis .
we finally note that our analysis also suggests that it is unlikely that sterile neutrinos could act in that role .
a part of the observations were made at the w.m .
keck observatory which is operated as a scientific partnership between the california institute of technology and the university of california ; it was made possible by the generous support of the w.m .
keck foundation .
this paper also includes data gathered with the 6.5 meter magellan telescopes located at las campanas observatory , chile .
the authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain .
the hydrodynamical simulations in this work were performed using the cosmos supercomputer in cambridge ( uk ) , which is sponsored by sgi , intel , hefce and the darwin supercomputer of the university of cambridge high performance computing service ( http://www.hpc.cam.ac.uk/ ) , provided by dell inc . using strategic research infrastructure funding from the higher education funding council for england .
cosmos and darwin are part of the dirac high performance computing facility funded by stfc .
mv is supported by the fp7 erc grant `` cosmoigm '' ga-257670 , prin - miur and infn / pd51 grants .
gdb acknowledges support from the kavli foundation .
jsb acknowledges the support of a royal society university research fellowship .
mh acknowledges support by the fp7 erc grant emergence-320596 .
the ratio of the 1d flux power spectrum for two wdm 2 kev simulations at different resolutions , ( 20,768 ) and ( 20,896 ) represented by the black and blue curves respectively , to the reference wdm 2 kev run ( 20,512 ) at two redshifts ( @xmath114 represented by the continuous and dot - dashed curves , respectively ) .
the mean flux is the same in all the models shown , and the shaded area indicates the range of wavenumbers used in the present analysis . ] in figure [ fig_e ] we compare the 1d flux power spectrum extracted from the wdm 2 kev @xmath62 and @xmath63 models to our reference resolution of @xmath53 .
we focus on this wdm mass since it is excluded at high significance by our analysis , yet it is used by the large scale structure community in order to solve apparent problems of @xmath8cdm at small scales .
there is at most a 10 - 12% correction at the smallest scales probed in our analysis ( @xmath244 s / km ) when the flux power spectra from the @xmath62 model are considered , and there is an extra 2% correction when the @xmath63 is considered . after correcting for this
, we therefore believe that we have reached 5% agreement in terms of the flux power for the smallest scales considered in this work .
this is below the 7.75% ( 1@xmath27 ) statistical error of the data at the same wavenumber . _
upper panel : _ the ratio of the 1d flux power spectrum for instrument resolutions corresponding to hires ( 6.7 kms@xmath32 ) and mike ( 13.6 kms@xmath32 ) spectrographs at 4 different redshifts ( @xmath49 ) to the data with the native resolution of the simulation .
_ lower panel : _ the ratio of the 1d flux power spectrum for a s / n value per pixel equal to 15 to a model s / n=@xmath245 .
the mock data have been corrected to incorporate both of these effects .
the mean flux is the same in all the models shown , and the shaded area indicates the range of wavenumbers used in the present analysis.,title="fig : " ] _ upper panel : _ the ratio of the 1d flux power spectrum for instrument resolutions corresponding to hires ( 6.7 kms@xmath32 ) and mike ( 13.6 kms@xmath32 ) spectrographs at 4 different redshifts ( @xmath49 ) to the data with the native resolution of the simulation . _
lower panel : _ the ratio of the 1d flux power spectrum for a s / n value per pixel equal to 15 to a model s / n=@xmath245 .
the mock data have been corrected to incorporate both of these effects .
the mean flux is the same in all the models shown , and the shaded area indicates the range of wavenumbers used in the present analysis.,title="fig : " ] in figure [ fig_h ] we show the effect that instrumental resolution and a given signal - to - noise ( s /
n ) ratio has on the lyman-@xmath0 flux power spectrum in the four different redshift bins of our data sample .
these results have been obtained from our mock qso spectra sample .
note that both of these effects have been incorporated into the lyman-@xmath0 forest spectra used in our analysis .
the s / n ratio results in an increase in the flux power of less than 5% over the range of wavenumbers considered in this work , while instrumental resolution effects are particularly important for the mike sample .
there is a 20% correction at the smallest scales for the mike data ( for the hires data this value is 5% ) .
[ [ systematic - effects - on - the - flux - power - induced - by - metals - and - uv - background - fluctuations ] ] systematic effects on the flux power induced by metals and uv background fluctuations ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ _ upper panel : _ the ratio of the 1d flux power spectrum for a model including metal line contamination to the reference model at four different redshifts . _
lower panel : _ the ratio of the 1d flux power spectrum for a model including the effect of spatial fluctuations of the uv background at four different redshifts .
note the different y - axis in the two figures .
the mean flux is the same in all the models shown , and the shaded area indicates the range of wavenumbers used in the present analysis.,title="fig : " ] _ upper panel : _ the ratio of the 1d flux power spectrum for a model including metal line contamination to the reference model at four different redshifts . _
lower panel : _ the ratio of the 1d flux power spectrum for a model including the effect of spatial fluctuations of the uv background at four different redshifts .
note the different y - axis in the two figures .
the mean flux is the same in all the models shown , and the shaded area indicates the range of wavenumbers used in the present analysis.,title="fig : " ] we also consider two important astrophysical nuissance effects in this work : unwanted contamination from metal lines in the lyman-@xmath0 forest and the effect of spatial fluctuations in the uv background intensity on the observed lyman-@xmath0 forest transmission . in the redshift range we consider in this work , @xmath99@xmath105 ,
the most common metal lines in the lyman-@xmath0 forest will arise from absorbers at lower redshifts .
we therefore consider the effect of absorption from three prominent absorption line doublets ; ( @xmath2461548 , 1551 ) , ( @xmath2461394 , 1403 ) and ( @xmath2462796 , 2804 ) , arising over the redshift intervals @xmath247@xmath248 , @xmath249@xmath250 and @xmath251@xmath252 , respectively .
we add these metal lines to lyman-@xmath0 forest spectra drawn from our reference @xmath8cdm hydrodynamical model using the following prodcedure .
we firstly integrate fits to the column density distribution functions ( cddfs ) presented by @xcite from a set of 19 high resolution vlt / uves quasar spectra at @xmath253@xmath254 over the column density range @xmath255-@xmath256 , for all three species .
we then multiply the results by the redshift path length of our lyman-@xmath0 forest data set to provide an estimate of the number of metal line absorbers in the lyman-@xmath0 forest .
note that this approach will likely overestimate the number of and absorbers due to the somewhat lower redshift coverage of the @xcite dataset relative to this work .
our metal contamination estimates are therefore likely to be conservative in this regard .
next , we monte carlo sample column densities and line widths for the appropriate number of lines from the @xcite cddf fits and the doppler parameter distribution given by ref .
@xcite , with a chosen value @xmath257 for all three species . finally , we randomly insert these absorption features into the sight - lines in our mock lyman-@xmath0 forest data set . in order to estimate the impact of spatial fluctuations in the uv background on the lyman-@xmath0 forest at @xmath99@xmath105 we use a modified version of the approach described in @xcite , which was used to examine fluctuations in the ionising background at lower redshift .
we refer the reader to ref .
@xcite for further details .
the key difference in this work is that we compute the spatially varying photo - ionisation rate along our simulated sight - lines .
we achieve this by computing the specific intensity of the ionising background between @xmath195@xmath258 ( replacing equation 3 in @xcite ) by solving : where @xmath260 is the frequency of the ionisation edge , @xmath261 is the mean free path for ionising photons presented by @xcite and @xmath262 is the power - law slope of the cddf @xcite .
the summation in eq .
[ eq1 ] is over all quasars with luminosities , @xmath263 , drawn from the @xcite b - band quasar luminosity function . in order to be conservative ,
in this work we adopt an extreme model which maximises the effect of the uv fluctuations on the lyman-@xmath0 forest by assuming all ionising photons in the igm at @xmath99@xmath105 are produced by quasars with @xmath264 .
we therefore ignore the significant contribution to the uv background from the more numerous , fainter star - forming galaxies at these redshifts , which effectively smooth out the large - scale fluctuations produced by the rarer quasars .
the spatially fluctuating photo - ionisation rates are then obtained by integrating the specific intensity with respect to frequency , weighted by the photo - ionisation cross - section . in figure [ fig_hh ]
we show the effect that uv fluctuations and metal contamination have on the flux power spectrum . in the wavenumber range considered here ,
the uv background fluctuations have an effect at around the 10% level at the largest scales , dropping to 5% at smallest scales considered in this work . the effect is larger at high redshift ( z=5,5.4 ) than in the two other redshift bins .
metal contamination affects the flux power very little ( at the @xmath265 level ) and is at a level below the statistical error bars of our data . the ratio of the 1d flux power spectrum for 2 different @xmath8cdm models ( @xmath266 in orange and @xmath267 in black ) at 4 different redshifts ( @xmath49 represented by the dot - dashed , dashed , dotted and continuous curves , respectively ) to the corresponding reference @xmath8cdm simulation . the wdm 2kev model is also shown in blue for comparison
the shaded area indicates the range of wavenumbers used in the present analysis . ] the mean flux level @xmath268 , or alternatively the effective optical depth @xmath269 , is a key ingredient in our flux modeling procedure and a quantity that needs to be marginalized over in the monte carlo markov chain likelihood estmation . in figure [ fig_gg ]
we demonstrate the effect that a different mean flux level has on the flux power spectrum .
we choose two different mean flux levels with @xmath84 20% higher and lower than the reference value ( which corresponds to the observed mean flux ) . a higher ( lower ) value for @xmath39 will result in more ( less ) power relative to the reference value .
the trends are similar to those found for the evolution of @xmath68 , i.e. rather flat in wavenumber space , with some weak scale - dependence which is only present in the highest and lowest redshift bins .
our final results are not sensitive to the actual choice of the effective optical depth values , since these are marginalized over in the likelihood estimation . p. bode , j. p. ostriker and n. turok , astrophys .
j. , 93 ( 2001 ) ; b. moore , t. quinn , f. governato , j. stadel and g. lake , mon . not .
* 310 * , 1147 ( 1999 ) ; v. avila - reese , p. colin , o. valenzuela , e. donghia and c. firmani , astrophys .
j. * 559 * , 516 ( 2001 ) ; m. miranda , macci a.v , mon . not .
* 366 * , 1529 ( 2006 ) .
boylan - kolchin , michael ; bullock , james s. ; kaplinghat , manoj , mon . not .
soc . * 422 * , 1203 ( 2012 ) ; pearrubia , jorge ; pontzen , andrew ; walker , matthew g. ; koposov , sergey e. , apj , * 759 * , 42 , ( 2012 ) .
lovell , mark r. ; eke , vincent ; frenk , carlos s. ; gao , liang ; jenkins , adrian ; theuns , tom ; wang , jie ; white , simon d. m. ; boyarsky , alexey ; ruchayskiy , oleg , mon . not .
soc . * 420 * , 2318 ( 2012 ) .
heitmann , katrin ; white , martin ; wagner , christian ; habib , salman ; higdon , david , the astrophysical journal , * 715 * , 1 ( 2010 ) ; semboloni , elisabetta ; hoekstra , henk ; schaye , joop ; van daalen , marcel p. ; mccarthy , ian g. , mon . not .
417 * , 2020 ( 2011 ) ; van daalen , marcel p. ; schaye , joop ; booth , c. m. ; dalla vecchia , claudio , mon . not .
soc . * 415 * , 4 ( 2011 ) busca , n. et al .
astronomy & astrophysics , * 552 * , 18 ( 2013 ) ; slosar , a. et al .
journal of cosmology and astroparticle physics , * 04 * , 026 , ( 2013 ) ; dawson , k. et al . , astronomical journal , * 145 * , 10 , ( 2013 ) ; ahn , c.p .
et al . , astrophysical journal supplement , * 203 * , 21 , ( 2012 ) .
a. boyarsky , a. neronov , o. ruchayskiy and m. shaposhnikov , phys . rev .
d * 74 * , 103506 ( 2006 ) ; c. r. watson , j. f. beacom , h. yuksel and t. p. walker , phys .
d * 74 * , 033009 ( 2006 ) ; a. boyarsky , j. w. den herder , a. neronov , o. ruchayskiy , astroparticle physics , 28 , 3 , ( 2007 ) ; abazajian k. , fuller g.m . and tucker w.h . ,
astrophys .
j. * 562 * , 593 ( 2001 ) ; k. abazajian and s. m. koushiappas , phys .
d * 74 * , 023527 ( 2006 ) macci , andrea v. ; ruchayskiy , oleg ; boyarsky , alexey ; munoz - cuartas , juan c. , mon . not .
soc . * 428 * , 882 ( 2013 ) ; anderhalden , d. ; diemand , j. ; bertone , g. ; macci a. v. ; schneider , a. ; journal of cosmology and astroparticle physics , * 10 * , 047 ( 2012 ) ; villaescusa - navarro , francisco ; dalal , neal , journal of cosmology and astroparticle physics , * 03 * , 024 ( 2011 ) macci , andrea v. ; fontanot , fabio , mon . not .
soc . * 404 * , 16 ( 2010 ) ; anderhalden , donnino ; schneider , aurel ; macci , andrea v. ; diemand , juerg ; bertone , gianfranco , journal of cosmology and astroparticle physics , * 03 * , 014 ( 2013 ) ; schneider , aurel ; smith , robert e. ; reed , darren , ( 2013 ) arxiv:1303.0839 benson , a.j .
et al . , mon . not .
* 428 * , 1774 ( 2013 ) ; angulo , raul e. ; hahn , oliver ; abel , tom , eprint arxiv:1304.2406 ; lapi , a. ; salucci , p. ; danese , l. eprint arxiv:1305.7382 ; fabio pacucci , andrei mesinger , zoltan haiman , eprint arxiv:1306.0009 | we present updated constraints on the free - streaming of warm dark matter ( wdm ) particles derived from an analysis of the lyman-@xmath0 flux power spectrum measured from high - resolution spectra of 25 @xmath1 quasars obtained with the keck high resolution echelle spectrometer ( hires ) and the magellan inamori kyocera echelle ( mike ) spectrograph .
we utilize a new suite of high - resolution hydrodynamical simulations that explore wdm masses of 1 , 2 and 4 kev ( assuming the wdm consists of thermal relics ) , along with different physically motivated thermal histories .
we carefully address different sources of systematic error that may affect our final results and perform an analysis of the lyman-@xmath0 flux power with conservative error estimates . by using a method that samples the multi - dimensional astrophysical and cosmological parameter space ,
we obtain a lower limit @xmath2 kev ( @xmath3 ) for warm dark matter particles in the form of early decoupled thermal relics . adding the sloan digital sky survey ( sdss ) lyman-@xmath0 flux power spectrum does not improve this limit .
thermal relics of masses 1 kev , 2 kev and 2.5 kev are disfavoured by the data at about the @xmath4 , @xmath5 and @xmath6 c.l .
, respectively .
our analysis disfavours wdm models where there is a suppression in the linear matter power spectrum at ( non - linear ) scales corresponding to @xmath7mpc which deviates more than 10% from a @xmath8cdm model .
given this limit , the corresponding `` free - streaming mass '' below which the mass function may be suppressed is @xmath9 m@xmath10 .
there is thus very little room for a contribution of the free - streaming of wdm to the solution of what has been termed the small scale crisis of cold dark matter . |
jets as highly collimated beams of high velocity material and outflows of comparatively lower degree of collimation and lower speed are an ubiquitous phenomenon among astrophysical objects .
jets are powerful signs of activity and are observed over a wide range of luminosity and spatial scale . among the jet sources are young stellar objects ( yso ) , micro - quasars , active galactic nuclei ( agn ) , and most probably also gamma ray bursts @xcite .
the common models of launching , acceleration , and collimation work in the framework of magnetohydrodynamic ( mhd ) forces ( see e.g. @xcite ) , although the details of the process are not fully understood . jets and outflows from yso and
agn affect their environment , and , thus , the formation process of the objects they are launching them .
numerous studies investigate effects of such feedback mechanisms in star formation and galaxy formation ( see e.g. @xcite ) .
however , a quantitative investigation of how much of mass , momentum , or energy from the infall is actually recycled into a high speed outflow needs to resolve the innermost jet - launching region and to model the physical process of launching directly .
this is the major aim of the present paper . according to the current understanding , accretion and ejection
are related to each other .
one efficient way to remove angular momentum from a disk is to connect it to a magnetized outflow .
this has been motivated by the observed correlation between signatures of accretion and ejection in jet sources ( see e.g. @xcite ) .
the overall idea is that the energy and angular momentum are extracted from the disk by an efficient magnetic torque relying on a global , i.e. large - scale magnetic field threading the disk .
if the inclination of the field lines is sufficiently small , magneto - centrifugal forces can accelerate the matter along the field line . beyond the alfvn point also lorentz forces contribute to the acceleration .
the collimation of the outflow is thought to be achieved by magnetic tension due to a toroidal component of the magnetic field .
still , we have to keep in mind the fact the the toroidal field pressure gradient acts de - collimating , and an existing external gas over - pressure may contribute to jet collimation . before going into further details , we like to make clear that with jet _ formation _ we denote the process of accelerating and collimating an already existing slow disk wind or stellar wind in to a jet beam . with jet _ launching _ we denote the process which conveys material from radial accretion into a vertical ejection , thereby lifting it from the disk plane into the corona , and thus establishing a disk wind . a vast literature exists on magnetohydrodynamics jet modeling .
we may distinguish i ) between steady - state models and time - dependent numerical simulations , or ii ) between simulations considering the jet formation only from a fixed - in - time disk surface and simulations considering also the launching process , thus taking into account disk and jet evolution together .
steady - state modeling have mostly followed the self - similar blandford & payne approach ( e.g. @xcite ) , but also fully two - dimensional models were proposed @xcite , some of them taking even into account the central stellar dipole @xcite .
further , some numerical solutions have been proposed by .@xcite in a weakly ionized protostellar accretion discs that are threaded by a large - scale magnetic field as a wind - driving accretion disk .
they have studied the effects of different regimes for ambipolar diffusion or hall and ohm diffusivity dominance in these disk .
( self - similar ) steady - state models have also been applied to the jet launching domain @xcite , connecting the collimating outflow with the accretion disk structure .
in addition to the steady - state approaches , the magneto - centrifugal jet formation mechanism has been subject of a number of time - dependent numerical studies . in particular ,
@xcite and @xcite have demonstrated for the first time the feasibility of the mhd self - collimation property of jets .
we note , however , that it was already 1985 when the first jet formation simulations were published , in that case for much shorter simulation time scale @xcite . among these works ,
some studies investigated artificial collimation @xcite , a more consistent disk boundary condition @xcite , the effect of magnetic diffusivity on collimation @xcite , or the impact of the disk magnetization profile on collimation @xcite . in the aforementioned studies ,
the jet - launching accretion disk is taken into account as a boundary condition , _ prescribing _ a certain mass flux or magnetic flux profile in the outflow .
this is a reasonable setup in order to investigate jet formation , i.e. the acceleration and collimation process of a jet .
however , such simulations can not tell the efficiency of mass loading or angular momentum loss from flux from disk to jet , or can not help answering the question which kind of disks do launch jets and under which circumstances .
it is therefore essential to extend the jet formation setup and include the launching process in the simulations .
numerical simulations of the mhd jet launching from accretion disks have been presented first by @xcite and @xcite , treating the ejection of a collimated outflow out of an evolving - in time resistive accretion disk .
@xcite further developed this approach with emphasis on how resistive affect the dynamical evolution .
an additional central stellar wind was considered by @xcite .
further studies were concerned about the effects of the absolute field strength or the field geometry , in particular investigating field strengths around and below equipartition @xcite .
these latter were _ long - term _
simulations for several 100s of ( inner ) disk orbital periods , providing sufficient time evolution to also reach a ( quasi ) steady state for the fast jet flow .
finally , we like to emphasize the fact that jets and outflows are observed as _ bipolar _ streams . jet and counter jet appear typically asymmetric with only very few exceptions .
one exception is the protostellar jet hh212 showing an almost perfectly symmetric bipolar structure @xcite .
so far , only very few numerical simulations investigating the bipolar launching of disk jets have been performed . among them
are the works of @xcite or @xcite which even included a disk dynamo action .
recent publications consider asymmetric ejections of stellar wind components from an offset multi - pole stellar magnetosphere @xcite .
it is therefore interesting to investigate the evolution of both hemispheres of a _
global _ jet - disk system in order to see whether and how a global asymmetry in the large - scale outflow can be governed by the disk evolution .
this has not been done so far . in a series of two papers
, we will address both the detailed physics of mhd jet launching ( paper i ) and the bipolarity aspects of jets ( paper ii ) . in the present paper ( paper i )
we investigate the details of the launching , acceleration , and collimation of mhd jets from resistive magnetized accretion disks .
we investigate how the mass and angular momentum fluxes depend on the internal disk physics applying long - lasting global mhd simulations of the disk - jet system .
we hereby investigate different resistivity profiles of the disk .
this paper is organized as follows .
section 2 is dedicated to mhd equations and to describe the numerical setup , the initial and boundary conditions of our simulations .
the general evolution of jet launching and the physical processes involved are presented in section 3 with the help of a reference simulation .
section 4 is then devoted to a parameter study comparing jets from different setups . in paper ii we will present the bipolar jets simulations , discussing their symmetry properties and how symmetry can be broken by the intrinsic disk evolution .
we model the launching of a mhd outflow from a slightly sub - keplerian disk , initially in pressure equilibrium with a non - rotating disk - corona . as illustrated in fig .
[ fig : cartoonshow ] , matter is first accreted radially along a disk surrounding a central object and is then loaded on to the magnetic field lines .
the large - scale magnetic field is threading the disk and thereby connects the accretion and ejection processes .
our main goals are * to determine the relevant mass fluxes ( mass ejection and accretion rate ) , and to study the influence on them by the leading physical parameters such as magnetic diffusivity , magnetic field strength , * to determine the resulting jet geometry - that is the asymptotic jet radius and opening angle , along with the size of the jet launching area of the disk , and the asymptotic jet velocity .
we will in a follow - up paper further extend our setup into two hemispheres , to investigate iii ) the launching of bipolar jets and their symmetry characteristics .
the relevant mass and energy fluxes for accretion and ejection are of essential interest for feedback mechanisms in star formation or galaxy formation . we our highly resolved simulations of the innermost regions of these objects we intend to quantify these properties for a range of possible parameters .
[ fig : cartoonshow ] for our numerical simulations , we apply the mhd code pluto 3.01 @xcite , solving the conservative , time - dependent , resistive , inviscous mhd equations , namely for the conservation of mass , momentum , and energy , @xmath1 @xmath2 + \nabla \left [ p + \frac{b^2}{8\pi } \right ] + \rho \nabla \phi = 0,\ ] ] @xmath3\\ = - \lambda_{\rm cool}. \end{gathered}\ ] ] here , @xmath4 is the mass density , @xmath5 the velocity , @xmath6 the thermal gas pressure , @xmath7 the magnetic field , and @xmath8 the gravitational potential of the central object of mass @xmath9 , withe the spherical radius @xmath10 . in general ,
the magnetic diffusivity is defined as a tensor @xmath11 ( see 2.5 ) . the evolution of the magnetic field is described by the induction equation , @xmath12 with the electric current density @xmath13 given by ampre s law @xmath14 .
the cooling term @xmath15 can be expressed in terms of ohmic heating @xmath16 , with @xmath17 , and with @xmath18 measuring the fraction of the magnetic energy that is radiated away instead of being dissipated locally . for simplicity ,
here we adopt @xmath19 .
the gas pressure follows an equation of state @xmath20 with the polytropic index @xmath21 and the internal energy density @xmath22 .
the total energy density is @xmath23 our simulations are performed in axisymmetry applying cylindrical coordinates .
the ceno3 algorithm as @xmath24 order interpolation scheme is used for spatial integration @xcite together with a @xmath24 order runge - kutta scheme for time evolution and an hll riemann solver . for the magnetic field evolution
we apply the constrained transport method ( fct ) ensuring solenodality @xmath25 . throughout the paper distances
are expressed in units of the inner disk radius @xmath26 , while @xmath27 and @xmath28 denote the disk pressure and density at this radius , respectively .
the index i refers to a number value at the inner disk radius at @xmath29 and time @xmath30 . in the appendix [ app : units - normalization ]
we show for comparison the astrophysical scaling for yound stellar ( yso ) jets and agn jets . typically , @xmath31 au for yso and @xmath32 schwarzschild radii for agns .
naturally , we can not treat any relativistic effects of agn jets with our non - relativistic setup .
velocities are measured in units of the keplerian speed @xmath33 at the inner disk radius .
time is measured in units of @xmath34 , which can be related to the keplerian orbital period @xmath35 .
pressure is given in units of @xmath36 .
the magnetic field is measured in units of @xmath37 . as usual
we define the aspect ratio of the disk @xmath38 as the ratio of the isothermal sound speed to the keplerian speed , both evaluated at disk mid plane , @xmath39 . ] . for a more details
see appendix [ app : units - normalization ] .
we define initial conditions following a setup applied by other authors before - a magnetically diffusive accretion disk is prescribed in sub - keplerian rotation , above which a hydrostatic corona in pressure balance with the disk is located @xcite .
the coronal density is chosen several orders of magnitude below the disk density , thus implying an entropy and a density jump from disk to corona . however , contrary to previous authors @xcite we do not apply an initial vertical velocity profile in the disk , and only prescribe a radial velocity profile .
we have seen that for our long - term simulations the whole disk system does adjust to a new dynamical equilibrium which does not depend on the vertical profile of the initial velocity distribution .
we prescribe an initially geometrically thin disk with @xmath40 which is itself in vertical equilibrium between thermal pressure and gravity .
we follow the standard setup employed in a number of previous papers @xcite . as initial disk density distribution
we prescribe @xmath41 \right)^{3/2}\ ] ] @xcite , while the initial disk pressure distribution follows @xmath42 the disk is set into slightly sub - keplerian rotation accounting for the radial gas pressure gradient and advection , @xmath43 following @xcite , but neglecting viscosity . in our setup the initial poloidal velocity is imposed by hand following the prescription of @xcite , @xmath44 simulations with initially vanishing disk accretion resulted in the same asymptotic inflow - outflow evolution . starting with a disk inflow as in eq .
[ eq : disk_vp ] , the inflow - outflow structure is established on a shorter time scale .
uniform grid cells on a physical domain of @xmath45 .
another 2000 stretched grid cells are attached in vertical direction @xmath46 . for the high resolution simulations a physical domain of @xmath47
is covered with @xmath48 grid cells to which another 2000 stretched grid cells are attached in vertical direction between @xmath49 . ] above the disk , we define an initially hydrostatic density and pressure stratification ( the so - called `` corona '' ) , @xmath50 the parameter @xmath51 quantifies the initial density contrast between disk and corona .
we adopt @xmath52 .
the initial magnetic field is prescribed by the magnetic flux function @xmath53 following @xcite , @xmath54 here @xmath55 measures the vertical field strength at @xmath56 .
the magnetic field components are calculated by @xmath57 and @xmath58 .
the computational domain spans a rectangular grid region applying a purely uniform spacing in radial direction and a uniform plus stretched spacing vertically ( fig .
[ fig : domain1 ] ) .
we have run simulations in two different resolutions ( see tab . [
tbl : cases ] ) . for the low resolution simulations the domain extends to @xmath59 inner disk radii @xmath26 on a grid of @xmath60 cells , resulting in a resolution of @xmath61 . for the high resolution simulations
the resolution is increased to @xmath62 on the cost of being limited to a smaller domain of @xmath63 .
for all simulations presented here , a uniform grid is used for the magnetically diffusive disk . for the typical disk model applied with @xmath64
the low resolution provides only 2 grid cells per disk scale height at the inner disk radius , while for somewhat larger radii , say at @xmath65 , we have about 10 cells per disk scale height . in the high resolution runs the disk resolution is increased by a factor 2.5 . as discussed by @xcite the low resolution barely resolves the jet launching disk surface layer of the very inner disk , where steep gradients in density , pressure and diffusivity are present . here , numerical diffusivity is supposed to play a role and we have devoted a section to investigate this effect . in order to follow the jet outflow over long distances and in order to provide a sufficient mass reservoir for disk accretion a large domain would be wishful . in order to resolve the wind launching area ,
a high resolution would be required . however , as mentioned before , pluto in its current version does not allow for diffusive mhd simulations in a stretched grid .
furthermore , we experiences that for large domains together with stretched grids ( implying an elongated shape of the outer grid cells ) the code had difficulties to solve the conservative equation and the simulations crashed . in the disk - jet system for the reference simulation ( see eq .
[ eq : magdiff ] ) . here , the diffusive scale height is @xmath66 . ] for the boundary conditions , axisymmetry on rotation axis and equatorial symmetry for the disk mid plane are imposed .
at the upper @xmath67 boundary , we use the standard pluto outflow boundary condition ( zero gradient ) , but at the outer radial boundary condition , we apply a modified outflow condition as derived by @xcite in order to avoid artificial collimation .
most essential is the internal boundary enclosing the origin which we call a sink .
we prescribe a sink for the mass flux in the very inner region of the domain .
the sink is essential for the following reasons .
first , the numerically problematic singularity at the origin can be hidden by this internal boundary .
second , the central sink allows to absorb the accretion flow through the disk and thus , emulate a central accreting object .
the sink is numerically introduced as an internal boundary condition , located on the region @xmath68 and @xmath69 .
is defined corresponding to the resolution , at least 4 cells is taken as a sink height in any run . ]
the internal boundary conditions defined at the right and top sides of the sink have significant impact on the simulation . if ill - defined , spurious effects arise during the early evolution already . for the right side of the sink ,
we impose _ accretion conditions _ , implying a zero gradient for pressure , density , and the vertical velocity component . for @xmath70 and @xmath71 a first order extrapolation ( constant gradient )
is imposed which ensures an angular momentum decrease across the boundary which is essential for accretion . to allow for accretion further , and also to minimize feedback from the sink to the domain , we constrain min@xmath72 in the ghost cells for the radial velocity component .
we assume that the magnetic flux is not advected into the central object , and we impose @xmath73
. the normal component of the magnetic field is calculated from the solenodality condition along both sides of the sink . on top of the sink
, we prescribe the initial local value for the gas pressure . in order to avoid evacuation of the regions close to the symmetry axis
, we impose a density of 110% the initial local density .
effectively , this condition replenishes the mass into the domain near to the axis to overcome numerically difficult low densities .
an essential point of our setup is to impose a proper outflow boundary at the outer boundary of the domain in @xmath74-direction avoiding artificial collimation forces .
here we have implemented a current - free outflow condition ( see appendix [ app : outer - boundary ] ) which avoids spurious collimation by lorentz forces and which has been thoroughly tested in our previous papers @xcite . in order to enable a long - term disk - jet simulation requiring a long - living disk accretion ,
we need to provide a sufficiently long lasting mass reservoir .
this could be realized by either i ) a prescribed mass inflow , by a ii ) local mass replenishment , or by iii ) providing a large mass reservoir for jet launching part of the accretion disk by extending the outer disk radius to large radii .
all these approaches have been chosen in the literature .
we decided to follow option iii ) and provide a sufficiently high mass reservoir outside of the inner launching region by extending the computational grid ( and thus the outer disk radius ) up to about 100 inner disk radii .
similar approaches haven been applied by @xcite .
we will see below ( see sect .
3.2 ) that our approach is working well , but has , however , its limits if applied for a very long lasting simulations . during about 5000 dynamical time steps
we loose about 30 - 40% of disk the mass due to ejection and accretion and due to unwanted mass loss across the outer disk boundary .
a dissipative effect is required for steady - state accretion in order to allow matter to diffuse across through the magnetic field threading the disk .
magnetic diffusivity allows the mass flux to cross the field lines and thus to allow for accretion along the disk .
we consider the magnetic diffusivity to be turbulent in nature , or `` anomalous '' , and , thus , can not be calculated self - consistently in our setup .
the origin of the turbulent magnetic diffusivity is usually referred to be as caused by the magneto - rotational instability ( mri ) @xcite in a moderately magnetized disk .
nevertheless , we argue that as we load turbulently diffusive disk plasma into the outflow , it is natural to assume that the outflowing material is initially turbulent initially as well .
therefore , we have the option of applying a diffusive scale height @xmath75 larger than the thermal scale height @xmath76 of the disk , supposing that the turbulent disk material lifted into the outflow will remain turbulent a few more scale lengths until turbulence decays .
it is clear that the strong magnetization in the upper layers of the disk will prevent generation of turbulence by e.g. the mri and will also quench the turbulence in the outflowing material . without going into detail
we may estimate the time scale for the decay of the turbulence potentially existing in the outflowing material as follows .
the decay of ( helical ) mhd turbulence follows a power law @xmath77 , where @xmath78 is the magnetic energy and the power law index depends on further details but is in the range of @xmath79 @xcite .
simulations indicate that mhd turbulence decays as fast as the hydrodynamic turbulence on a few eddy turnover times @xcite . with @xmath80 ( in code units )
, we find a time scale for the decay of mhd turbulence launched from the disk into the outflow of e.g. @xmath81 , thus in the range of half an orbital period .
this must be compared to the kinematic time scale for the jet launching .
if we consider the propagation of the initially slow disk wind over a few disk pressure scale heights @xmath82 , we find that this time is of the order of a few eddy turnover times . also , we will later see that jet launching is a rapid process with an outflow being established already after a few dynamical time steps .
given the essential role of the magnetic diffusivity for wind / jet launching , we decided to investigate the launching process for different strength and different scale height of diffusivity .
we have run simulations for a variety of combinations of parameter values , but for sake of comparison , in this paper , we show results for diffusive scale height values , @xmath83 .
formally , we apply an @xmath0 prescription @xcite , similar also to previous works ( see e.g. @xcite ) .
we assume the diffusivity tensor to be diagonal with the non - zero components @xmath84 here , we have defined a poloidal magnetic diffusivity @xmath85 , and the toroidal magnetic diffusivity @xmath86 , respectively effectively governs the diffusion of the poloidal magnetic field , while the @xmath87 and @xmath88 govern the diffusion of the toroidal magnetic field ] , with a function @xmath89 describing the diffusivity profile . as it is known from the literature ,
an _ an - isotropic _ magnetic diffusivity is required to obtain stationary solutions @xcite .
most of the numerical simulations followed that approach and did find a stationary - state from their simulations as well ( see e.g. @xcite ) .
note that simulations of @xcite apply an isotropic diffusivity and reach a steady state as well .
we define an anisotropy parameter by the ratio of the toroidal to poloidal diffusivity components , @xmath90 . for the diffusivity function
@xmath91 several options can be considered . for the simulations in this paper ( paper i )
we apply @xmath92 where @xmath93 and @xmath94 govern the strength of magnetic diffusivity .
the alfvn speed @xmath95 , and the disk thermal scale height @xmath96 are both calculated along the mid - plane .
the parameter @xmath97 measures the scale height of the disk magnetic diffusivity , similar to the hydrodynamic disk scale height parameter @xmath98 .
for the present paper , we decided not to evolve the diffusivity profile in time .
we find that since both the disk scale height @xmath99 and the alfvn speed @xmath100 do vary as the disk evolves , the strength of diffusivity may vary substantially from the initially prescribed value ( up to a factor 10 ) .
this highly non - linear feedback may disturb the progress of our simulation such that an artificially high diffusivity may be derived which will affect the numerical time stepping .
a constant - in - time prescription of diffusivity simplifies our aim of disentangling the governing physical processes involved in jet launching . in the follow - up paper ii presenting truly bipolar jet launching
we will discuss further models for the magnetic diffusivity .
we again stress the importance to provide the reader with the actual number values of the magnetic diffusivity applied in time , only the initial diffusivity distribution was provided , butvnot its temporal evolution , and , thus , the magnetic diffusivity distribution which is actually involved in producing the outflow ( see e.g. @xcite ) .
see also our numerical example below . ] .
if we consider a @xmath101 together with a maximum alfvn speed in the disk of about @xmath102 ( located at the inner disk radius ) , this gives a maximum disk diffusivity of about @xmath103 , through all the simulations .
figure [ fig : diff_prof ] shows the distribution of the magnetic diffusivity .
the diffusivity is mainly concentrated in the disk , increases with disk radius ( consistent with a decreasing temperature or ionization degree with radius ) , and decreases with height resembling the transition from a cool disk to an ideal mhd outflow .
the simulations are governed by a set of non - dimensional parameters for the following physical quantities , * the magnetic field strength defined by plasma-@xmath104 , and its initial geometry defined by the `` bending '' parameter @xmath105 , * the magnetic diffusivity , with the three parameters @xmath94 , @xmath106 , @xmath107 governing its strength and anisotropy , and the diffusivity scale height , * the initial density contrast between disk and corona @xmath108 , * the initial disk thermal scale height parameter @xmath109 . an overview over these parameters are shown in tab .
[ tbl : cases ] ( first half ) for the simulations presented in this paper . the second part of tab .
[ tbl : cases ] shows the main dynamical quantities resulting from our simulations and will be discussed below .
table [ tbl : scale ] show the similar numbers for the simulations of different diffusive scale height .
before presenting detailed results of our parameter study of jet launching conditions , we will first discuss the general physical processes involved considering our long - term reference simulation ( case 1 ) .
this simulation will then later be compared to simulations applying i ) different diffusivity profiles , ii ) a higher grid resolution , iii ) and different magnetic field strength ( see tab . [
tbl : cases ] ) . a similar setup will then be used to launch bi - directional outflows , without the constraint on hemispheric symmetry ( see paper ii ) .
our reference simulation , case 1 is carried out up to @xmath110 dynamical time , corresponding to 796 inner disk orbital periods and is thus one of the longest simulations considering the mhd jet launching .
this huge time period corresponds , however , only to 3 rotations at a radius @xmath111 and correspondingly less at the outer grid radius @xmath112 .
consequently , while the inner part of the disk , and thus the outflow evolving from that part , has reached a steady state situation , the outer disk corona is still dynamically evolving .
we will therefore concentrate our discussion mainly to the inner disk areas .
the time evolution of density together with the magnetic field is shown in fig .
[ fig : rhocase1 ] . note that we do _ not _ show field lines integrated from certain foot point radii , but _ magnetic flux _
this allows to visualize the diffusion and advection of the magnetic field as a consequence of the disk evolution .
the magnetic diffusivity distribution determines the coupling between the plasma and magnetic field , and , thus , affects the mass loading into the outflow .
since the high diffusivity in the outer part of the disk , the coupling is weaker , and , thus , the mass loading less efficient . as a general result
we observe a continuous and robust outflow launched from the inner part of the disk , expanding into a collimated jet .
figure [ fig : poloida - velocity - vecplot ] shows the poloidal velocity distribution in jet and disk and the normalized poloidal velocity vectors indicating the direction of the mass fluxes . for reference
run case 1 .
shown is @xmath113 distribution in logarithmic scale overlaid with which arrows of the normalized poloidal velocity vectors indicating the direction of the flow , @xmath114 . ]
the outflow is accelerated to super fast magneto - sonic speed ( see the magnetosonic surfaces indicated in fig.[fig : rhocase1 ] ) a bow shock develops at the interface between the outflow and the surrounding hydrostatic corona . as the outflow propagates ,
the initial corona is swept out of the computational domain together with the bow shock .
while the interaction with the ambient gas plays a role within the first evolutionary steps , the long - term evolution of the outflow - its acceleration and collimation - is purely determined by the force balance within the outflow and the physical conditions of the launching region . as jets are launched from disks ,
the disk evolution is itself an essential part of the jet evolution and must be carefully considered .
it is expected that the jet mass flux would somewhat correspond to the disk accretion rate ( which would also depend on actual mass of the disk ) . in the disk for reference run ( case 1 ) for the dynamical times @xmath115.,title="fig : " ] in the disk for reference run ( case 1 ) for the dynamical times @xmath115.,title="fig : " ] in the disk for reference run ( case 1 ) for the dynamical times @xmath115.,title="fig : " ] in the disk for reference run ( case 1 ) for the dynamical times @xmath115.,title="fig : " ] in the disk for reference run ( case 1 ) for the dynamical times @xmath115.,title="fig : " ] our simulations show that the disk evolution is complex , with smooth accretion phases are followed by more turbulent stages .
the time evolution of the accretion stream @xmath116 is shown in fig .
[ fig : accretion_layer ] .
we see that accretion starts first at small radii and fully establishes after @xmath117 .
accretion shocks and vortices may destroy the quasi - steady state situation .
close to the outer disk radius , mass is lost through the outer boundary . at late evolutionary stages
we see a complex pattern in the @xmath116 distribution indicating small shock and turbulent motion within the disk . nevertheless , the negative mass fluxes in fig .
[ fig : accretion_layer ] , together with the negative velocity vectors in fig .
[ fig : poloida - velocity - vecplot ] clearly indicate that accretion is established throughout the whole disk .
we now discuss the time evolution of the disk mass for our reference simulation . in order to measure the disk mass ,
we need to carefully define the control volume defining a corresponding _ `` disk surface''_. here , we consider as disk surface the location where the mass density at each radius falls bellow @xmath118 of the mid - plane density ( because of the vortex motion we could not use the negative velocity as proxy for accretion ) .
measuring the mass contained in subsequent disk rings @xmath119 , we see that most of the mass is indeed stored within the outer disk areas ( fig .
[ fig : accelement ] , top ) .
we may identify two essential effects affecting the long - term disk evolution .
firstly , the mass reservoir in the outer disk has decreased - mainly due to outwards mass loss across the outer disk boundary .
secondly , the mass of the inner jet - launching disk has also decreased - due to both outflow activity and accretion into the sink . comparing the mass content of the disk at the initial and final stage , we find that the total disk mass decreases from @xmath120 at @xmath30 to @xmath121 at @xmath122 ( in code units ) . integrating the mass loss by accretion and ejection into the jet ( see fig .
[ fig : accelement ] ) over 5000 dynamical time steps , we find a total mass loss of @xmath123 .
the difference , i.e. @xmath124 , is the mass which is lost from the outermost disk in vertical and radial direction . in other words , a substantial amount of the disk mass loss is lost from the very outer part of the disk and does not directly influences the jet formation .
the disk mass which is lost from the inner part is partly lost as disk wind / jet , partly accreted into the sink and partly replenished by accretion from outer radii .
the time evolution of the disk mass gives a similar picture ( fig .
[ fig : accelement ] , bottom ) .
integrated over the whole domain , the disk loses about @xmath125 of its mass until @xmath110 in the reference simulation case 1 .
if we decrease the integration domain ( outer radius @xmath126 ) the inner disk looses less mass while part of its mass is being accreted from outer disk radii . in general
this implies that up to time @xmath122 , there is i ) still sufficient mass to support the accretion process , but also that ii ) the disk mass shows a considerable decrease which may have changed the internal disk properties .
thus , for an even longer - lasting simulation one would have to invent another mass source for the disk accretion ( e.g. by a physical mass inflow boundary condition at the outer disk radius properly taking into account angular momentum transfer , or a floor density distribution in the disk ) .
( integrated mass per disk ring , in code units ) at times @xmath127 ( top ) , and the time evolution of the total disk mass @xmath128 normalized to the initial disk mass @xmath129 for integration radii @xmath130 ( bottom).,title="fig : " ] ( integrated mass per disk ring , in code units ) at times @xmath127 ( top ) , and the time evolution of the total disk mass @xmath128 normalized to the initial disk mass @xmath129 for integration radii @xmath130 ( bottom).,title="fig : " ] along with the hydrodynamic disk evolution , the magnetic field distribution evolves as well , subject to the competing processes of advection and diffusion .
this also changes the launching conditions , as the overall profiles of plasma-@xmath104 , resp .
magnetization , are affected .
figure [ fig : adve_diff ] displays the time evolution of the magnetic flux surface @xmath131 for reference simulation case 1 , initially rooted at the radius @xmath132 .
we see that this flux surface is first moving ( diffusing ) outwards " driven " by magnetic pressure gradient and tension , and then , once disk accretion has established again moves inwards being advected with the mass flux .
after about @xmath133 both processes balance and a quasi steady - state situation in the system evolution is reached .
considering simply flux conservation , we may estimate the change in poloidal magnetic field strength and the corresponding change in magnetic diffusivity and plasma-@xmath104 .
estimating an average field strength @xmath134 and a corresponding magnetic flux @xmath135 , flux conservation tells us that @xmath136 .
therefore , @xmath137 , since this flux surface ( e.g. @xmath131 ) is now rooted at a different radius . with the increased poloidal magnetic field strength the alfvn speed @xmath138 increases , and , thus , also the magnetic diffusivity parameterized as @xmath139 .
thus , we expect that in previous work ( e.g. @xcite ) the actual ( time - dependent ) value of the magnetic diffusivity may differ substantially from its initial value .
we believe that this will impact the derived mass fluxes .
a similar estimate can be made for the plasma-@xmath104 or magnetization . since @xmath140 ,
the increase in plasma-@xmath104 is by a factor 100 , which also is expected to affect the jet formation severely .
we will come back to this point later when we compare different parameter runs .
the main goal of this paper is to investigate what fraction of the mass accretion is loaded into the outflow , and how the mass loading is affected by various disk parameters .
the derived mass ejection - to - accretion ratio can be an important ingredient for studies of agn or yso feedback , or another similar self - regulating outflow scenario .
efficient accretion is feasible only if angular momentum is sufficiently removed . since we do not consider a physical viscosity in our simulations , angular momentum removal from the disk
is mostly accomplished by the torque of the magnetized disk wind .
therefore , in our simulations disk accretion can only work if a disk wind has been established . in the following
we discuss the inflow and outflow rates in our disk - jet simulations , concentrating first on reference run ( case 1 ) . in order to calculate the integrated properties of inflow and outflow , such as mass flux or angular momentum flux , we need to carefully select the integration domain .
the derived fluxes depend strongly on the choice of the integration boundaries - thus , on how we distinguish between material belonging to _ accretion _ or _
ejection_. in reality there is no _ disk surface _ , but a smooth transition between accretion and ejection .
the initial setup of a thin disk with aspect ratio @xmath64 dynamically evolves in time , and so does the disk surface . and @xmath141 along the disk .
the integration surface denoted by @xmath142 is inclined and is parallel to the initial disk surface . ]
our control volume to measure the disk accretion and ejection rates is defined by an axisymmetric sector enclosed by two surfaces perpendicular to the equatorial plane located at @xmath143 and @xmath144 , and two other surfaces being the disk mid plane , and a surface @xmath145 which is close by and parallel to the initial disk surface ( see fig . [
fig : cv ] ) .
we usually adopt @xmath146 .
the accretion rate is then calculated as @xmath147 where the parameter @xmath148 controls the scale height of the integration , and @xmath149 is the initial thermal scale height of the disk .
considering the evolution of the large - scale disk velocity ( see fig .
[ fig : accretion_layer ] ) , we have chosen @xmath150 for our reference simulation . similarly , we calculate the ejected mass flux as @xmath151 where the integration is done along the inclined surface area s from @xmath152 to @xmath141 considering the area element @xmath153 . the time evolution of the accretion and ejection rate for the reference simulation is shown in fig . [ fig : accr_plot ] .
we have calculated the accretion and ejection rates for different sizes of the integration domain ( thus with different outer radius @xmath141 ) .
we observe that for small radii the accretion rate saturates at an approximately constant value .
for the outer parts of the disk , however , a longer dynamical time is required for the disk to evolve into a new dynamical steady state .
this is visible in the time evolution of the accretion rates - for increasing radii @xmath154 , the time when the plateau phase is reached , is increasing to @xmath155 . beyond the plateau phase
, the mass fluxes slightly decrease , most probably due to the overall decrease of the disk mass itself , and the subsequent change in the internal disk dynamics .
the accretion rates at large radii are larger than those at smaller radii - the mass difference is ejected as outflow . for the control volume with larger @xmath141 the ejection rate increases , which is simply a consequence of the increased integration area .
the last panel in fig .
[ fig : accr_plot ] shows the ejection - to - accretion rate @xmath156 .
again , depending on the size of the control volume this fraction changes .
although the outflow quickly evolves from the disk surface , and extends soon to large radii , several orbital periods are required to establish full accretion . at earlier times and for large radii ,
the accretion process is not fully established , resulting in a somewhat arbitrarily high ejection - to - accretion ratios even above unity .
for the inner part of the disk within radii @xmath157 , about 60% of the accreting material is diverted into the outflow for our reference simulation .
this is a large fraction and similar to simulations in the literature , but definitely more than derived in steady state models @xcite . for the other cases we have investigated ,
also smaller rates were obtained ( see below ) . .
, title="fig : " ] .
, title="fig : " ] .
, title="fig : " ] applying a radially self - similar approximation of the mhd equations , @xcite have introduced a so - called _ ejection index _ @xmath158 , @xmath159 essentially resulting from local mass conservation ( where @xmath160 and @xmath161 in our notation ) . with self - similar solutions @xcite constrained the ejection index to @xmath162 . in comparison , for our reference
run we find both larger numerical values and also a wider range for the ejection index , @xmath163 . table [ tbl : cases ] provides an overview over the ejection indexes we measure .
we can apply our reference run to different astrophysical sources ( see [ app : units - normalization ] ) , thus for a stellar jet with the central mass @xmath164 and @xmath165 , the accretion rate is @xmath166 .
this value for agn with the central mass @xmath167 and @xmath168 is about @xmath169 . in order to investigate the launching , the acceleration and the collimation processes of the outflow
, we now consider the forces acting in the disk - outflow system . to identify the forces for _ acceleration and collimation _ explicitly
, it is preferable to project them along or perpendicular to a certain flux surface , respectively . by comparing these projected force components
, we may disentangle the dominant driving and collimation mechanism .
figure [ fig : forces_case1 ] shows the force components along a flux surface ( field line ) rooted at @xmath170 , and also the critical mhd surfaces - the slow - magneto - sonic , the alfvn , and the fast - magneto - sonic surface .
the forces involved in _ driving the outflow _ are the centrifugal force @xmath171 , the lorentz force @xmath172 , the gas pressure gradient @xmath173 , and gravity @xmath174 . among them , the pressure gradient and the centrifugal force are de - collimating while gravity and lorentz force have collimating components .
the pressure gradient , centrifugal force and lorentz force also contribute to accelerate the outflow . in agreement with the literature
we see that upstream the slow magneto - sonic point , the acceleration is mainly by the gas pressure gradient . for the super - slow outflow the lorentz force and the centrifugal force dominate . for the super - fast flow the lorentz force plays a major role . in case of the collimating forces
the situation is different . before the alfvn point
, the de - collimating centrifugal force dominates . beyond the alfvn point
the lorentz force has a comparable contribution , and finally after the fast - magneto - sonic point , the lorentz force dominates and collimates the outflow .
the lorentz force behavior is a sign of the electric current distribution in the disk / jet system .
it compresses the disk while is collimating the corona . along the distance @xmath175 of the field line rooted at ( 5,0 ) . here
@xmath176 denote the centrifugal force , gravity , the lorentz and the gas pressure gradient forces ( all in absolute value ) , respectively .
the vertical lines present the slow - magnetosonic ( dashed line ) , the alfvn ( solid ) , and the fast - magnetosonic ( dot - dashed ) point , respectively.,title="fig : " ] along the distance @xmath175 of the field line rooted at ( 5,0 ) . here
@xmath176 denote the centrifugal force , gravity , the lorentz and the gas pressure gradient forces ( all in absolute value ) , respectively .
the vertical lines present the slow - magnetosonic ( dashed line ) , the alfvn ( solid ) , and the fast - magnetosonic ( dot - dashed ) point , respectively.,title="fig : " ] the forces involved in _ launching the outflow _ are mainly the vertical gas pressure gradient which counteracts the tension forces of the poloidal field and gravity ( see fig . [
fig : forces_case1 ] ) .
the thermal pressure gradient is always positive , and thus supports launching .
it increases from the disk mid plane to the surface and then decreases in the corona .
the tension term is mostly negative , thus compressing the disk , and does not support launching . the magnetic pressure ( toroidal component and total )
are both positive above the disk , and , thus , may support the launching process .
the toroidal component of the lorentz force @xmath177 provides the magnetic torque braking the disk , and loading the outflow with both angular momentum and energy ( not shown ) . in order to brake the disk ,
the torque must be negative in the disk and change sign at the disk surface @xcite .
this is confirmed by our simulation .
figure [ fig : launching - forces2 ] show for comparison the three main force components ( top ) and the net force ( bottom ) .
we see that in the disk the _ vertical force components _ almost balance , while above the disk a considerable net force remains which launches and accelerates the outflow . in the disk
the ( positive ) gas pressure is almost balanced by the ( negative ) lorentz force .
however , the gas pressure slightly wins and counteracts gravity .
essentially the figure shows that launching is a process which happens at the disk surface .
the outflow material is not lifted from the mid plane into the disk wind .
it is the disk material which is accreting along the disk surface which is loaded into the outflow .
the net ( specific ) force which is responsible for launching and initial acceleration is about 10% of the value of the single force components . at a radius @xmath65 for in case 1.,title="fig : " ] at a radius @xmath65 for in case 1.,title="fig : " ] [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^ " , ] [ tbl : scale ] we have argued above that the disk material which is launched to form an outflow is likely to be turbulent , so we may expect a magnetically diffusive disk wind above the disk surface .
thus , since mass loading requires poloidal magnetic diffusivity , the jet launching area can be extended into higher altitudes above the disk surface .
furthermore , the jet launching area is extended into a domain where the plasma-@xmath104 is comparatively low .
taking into account both effects , one may therefore expect i ) to launch jets from disks which would themselves be insufficiently magnetized for jet - driving , and also ii ) to launch outflow which have a rather low mass flux and subsequently higher terminal velocity . in order to investigate this effects ,
we have therefore prescribed different height scales @xmath178 for the magnetic diffusivity in our simulations ( see eq.[eq : magdiff ] ) .
the physical properties measured for different diffusive scale heights are shown in tab .
[ tbl : scale ] . in general
, we find different stages in the temporal evolution of the mass flux ( accretion and ejection ) in the disk - jet system ( fig .
[ fig : hstudy_scale ] ) . for small diffusive scale heights
@xmath178 , a turbulent pattern and perturbations appear in the outflow resulting in a more filamentary outflow .
it seems that these perturbation have larger amplitudes in case of a smaller diffusive scale height .
the amplitudes decrease when the diffusive scale height decreases .
for @xmath179 they are damped completely .
we believe that these perturbations are generated by a physical process within the launching region and that they are simply smoothed out by the magnetic diffusivity . a further study is needed to clarify this issue . for increasing scale height the accretion rate and ejection rate decrease , confirming the correlation derived above between the increasing magnitude of diffusivity and decreasing mass fluxes . in other words ,
an increasing diffusivity scale height simply corresponds to a higher diffusivity with the consequences as discussed above .
interestingly , all our simulations for different diffusivity scale heights converge to the same kinetic angular momentum flux in the outflow . on the other hand
, we find a trend of decreasing magnetic angular momentum flux with decreasing diffusivity scale height .
in this section we highlight some of our results discussing them in the context of numerical studies published previously .
our main goal was to perform a consistent parameter study to investigate a variety of physical effects involved in jet launching by using _
one _ code with _
one _ numerical setup .
only this allows us to disentangle the leading effects involved in jet launching without the uncertainty introduced by interpreting results obtained with different codes ( vac , pluto , flash ) or setups .
we first stress the point that we apply a well defined , fixed - in - time _ magnetic diffusivity _ distribution .
the magnetic diffusivity is an leading parameter which influences the whole disk - jet evolution @xcite .
previous simulations @xcite have applied a diffusivity evolving in time . this might be considered as more advanced , however , the previous authors do not provide details about time - evolution of the diffusivity , thus intruducing some level of uncertainty . here , we have demonstrated how diffusion and advection of magnetic flux does change the disk alfvn speed ( and thus the magnetic diffusivity if a time - dependent @xmath180 is applied ) by a factor of ten . on the other hand , by applying different strength and scale height for the diffusivity we investigate the interrelation between magnetic diffusivity and the accretion and ejection mass fluxes .
similarly , we provide information about the spatio - temporal _ evolution of the magnetization _ ( or plasma-@xmath104 in our case ) . @xcite have published a magnetization study , however , on a smaller grid and for substantially shorter dynamical evolution . in our paper
we show in particular how the magnetization profile flares over 1000s of dynamical time scales ( fig .
[ fig : log_beta_case1 ] ) . thus , along with the evolving outflow , the magnetization above the disk surface is substantially lower compared to the initial condition .
we obtain _ jet velocities _ which seem to be smaller compared to previously published values .
for example , @xcite detect asymptotic velocities of 1.5 - 1.8 times the footpoint keplerian speed for the field lines crossing the upper boundary .
simulations of jet formation from a fixed - in - time disk surface boundary ( see e.g. @xcite ) give similar values .
in contrast , we report lower velocities of about 0.8 times the keplerian speed at the inner disk radius .
note however , that our values result from a differently defined measure of the jet velocity .
we suggest that this mass - flux weigthed asymptotic speed - the average velocity of the bulk mass flux - is more applicable to observations .
the _ maximum jet velocities _ we observe in our simulations are higher and also up to 1.8 time the keplerian speed at the inner disk radius , which is consistent with the previous studies .
we explicitely consider the _ jet rotation _ in our simulations a topic which has not been treated in previous launching simulations .
we quantify the _ jet collimation _ degree by the directed mass fluxes . for the jet opening angle
we find somewhat larger values than in previous studies .
we suppose that this results from the modified outflow condition we have applied as we have seen that the original outflow condition lead to smaller opening angles .
we have run _ long - lasting simulations _ of up to 5000 dynamical time - scales @xmath181 .
this allows us to i ) reach a quasi steady - state situation of the simulation , but also to investigate the evolution beyond .
we find a slight but persistent change in the inflow - outflow dynamics on this very long time scales , such that the accretion rate slightly decreases .
the main reason for this we see in the decrease of the disk mass for such long time scales .
however , the effect has an direct astrophysical application as it may be applied to the long - term evolution of classical tts when the disk accretion in fact weakens after some @xmath182 years .
we note that previous simulations were substantially shorter , lasting until 400 dynamical time steps @xmath181 @xcite or 30 inner disk rotations @xcite .
an exception is the simulation by @xcite lasting for more than 900 inner disk orbits . that paper focusses , however , on the launching physics and does not show the large - scale evolution of the outflow .
together with the long time evolution , we have also have applied a _ large grid size_. compared to @xcite who investigate how magnetization affects launching by using a @xmath183 grid , our grid has about double size .
@xcite concentrate on the launching process of weakly magnetized disk and display only results on @xmath184 images ( which are a subsets of a @xmath185 ) .
for yso , our grid extension corresponds to about 28au along the jet , well into the observable region .
we note that we have applied an equidistant grid for the magnetically diffusive disk area which is consistent with the pluto code requirements .
this is different to @xcite who attach a scaled grid for the diffusive outer disk .
we have presented results of mhd simulations investigating the launching of jets and outflows from a magnetically diffusive disk in keplerian rotation .
the time evolution of the accretion disk structure is self - consistently taken into account .
the simulations are performed in axisymmetry applying the mhd code pluto .
the main goal of our simulations was to study how magnetic diffusivity ( its magnitude and distribution ) and magnetization affect the disk and outflow properties , such as mass and angular momentum fluxes , jet collimation , or jet radius .
our grids extend to @xmath59 inner disk radii with a resolution of @xmath186 , respectively @xmath187 inner disk radii with a higher resolution of @xmath188 .
an internal boundary ( sink ) is placed close to the origin absorbing the accreted mass and angular momentum .
we have prescribed a magnetic diffusivity in the disk based on an @xmath0-prescription .
one of our parameters was the scale height of the magnetic diffusivity with the option to have it higher than the thermal scale height .
this can be motivated considering that it is the turbulent disk material which is loaded into the outflow , and that the turbulence pattern is swept along with the disk wind until it decays .
we have investigated disks carrying a magnetic flux corresponding to an initial plasma - beta ranging from 10 to 5000 at the inner disk radius . as a general result
we observe a continuous and robust outflow launched from the inner part of the disk , expanding into a collimated jet and is accelerated to super fast magneto - sonic speed .
the key results of our simulations can be summarized as follows .
\(1 ) concerning the _ acceleration of the outflows _ , our simulations confirm that the magneto - centrifugal acceleration mechanism is most efficient in the low plasma-@xmath104 regime , while for weak magnetic fields the toroidal magnetic pressure gradient drives the ejected material .
we also confirm that the magnetocentrifugal mechanism also depends on the mass load in the outflow , as this mechanism works more efficiently for outflows with low mass fluxes .
however , compared to the magnetic pressure driven outflows , jets in the magnetocentrifugal acceleration regime have usually higher mass fluxes . in our simulations with very high plasma-@xmath104
we detect a highly unsteady behaviour \(2 ) efficient magneto - centrifugal driving which can accelerate jets to high kinetic energy relies on a strong coupling between magnetic field and the rotating disk , thus on a low diffusivity .
we find that it is the poloidal diffusivity @xmath189 which mainly affects the driving of the outflow .
however , besides the coupling needed for acceleration , also the _ launching of material _ depends on diffusivity . with increasing @xmath189 , the mass fluxes ( both the accretion rate and the ejection rate ) decrease
subsequently , the higher ejection rates result in a lower asymptotic outflow velocities .
\(3 ) we measure typical _ outflow velocities _ are in the range of @xmath190 times the inner disk rotational velocity with the tendency that the mass fluxes obtained in magneto - centrifugally driven outflow are substantially higher . here
, we confirm the clear ( inverse ) correlation between jet velocity and mass load , as it is well known from the literature .
note that we have apply a mass - flux - weighted jet velocity which we find more applicable to the observations . for the bulk mass flux
we find lower velocities compared to other papers , which are mostly dealing with the maximum speed obtained in the simulation .
our maximum velocities are similar . the relatively high speed of the outflows with low poynting flux which are driven by poloidal pressure gradient is unclear .
we find that the toroidal diffusivity affects the _ outflow rotation _ - a small toroidal diffusivity implies a larger jet rotational velocity .
this has not shown before in simulations .
\(4 ) we do not find a clear correlation between the _ outflow collimation _ and the magnetic field strength .
also weakly magnetized outflows , which are driven by the magnetic pressure gradient , and which we find to be quite unsteady , show a high degree of collimation .
the question of collimation for the weakly magnetized outflows is not answered .
we find that outflows within the magneto - centrifugal - driving regime the flows ejected from a weakly diffusive disks are only weakly collimated .
similarly , their jet radius ( here defined as mass flux weighted radius ) is larger in case of a lower poloidal magnetic diffusivity . follwoing the magnetic flux surface along the bulk mass flux from the asymtotic regime to the launching area
, we can defined the _ launching area _ of the outflow .
we find a size of the launching area from which the bulk of the mass flux originates in the range between 3 and 8 inner disk radii or about 0.4 au .
\(5 ) depending on the strength of magnetic diffusivity , the disk - jet structure may evolve into a steady - state .
we found that the cases with the strong field with @xmath191 and poloidal diffusivity @xmath192 will reach a quasi steady state , confirming the literature .
\(6 ) the magnetic flux profile along the disk is subject to advection and diffusion .
we find that the magnetization ( or plasma-@xmath104 ) of disk and outflow may therefore substantially change during the time evolution .
we have observed that the initial disk magnetization may change by a factor of 100 .
this may have severe impact on the launching process and the formation of the outflow in the sense that a rather highly magnetized disk may evolve into a weakly magnetized disk which can not drive strong outflows .
this issue has not been discussed before in the literature .
\(7 ) for very long time scales the accretion disk changes its internal dynamics , as due to outflow ejection and disk accretion the disk mass decreases . as a consequence ,
the accretion and ejection rates slightly decrease . in order to compensate for this effect
, we have applied a large outer disk radius providing a large mass reservoir for the inner jet - launching disk .
\(8 ) for our simulations , we found that 10 - 50% of the accreting plasma can be diverted into the outflow . for i )
less diffusive disks , ii ) a strong magnetic field , iii ) a low poloidal diffusivity , or a iv ) lower numerical diffusivity ( resolution ) , the mass loading into the outflow is increased - resulting in more massive jets .
we interpret as physical reason the more efficient extraction of angular momentum from the disk , due to the stronger matter - field coupling .
note that we do not consider in our simulations viscosity or the wealth of thermal effects which play an essential role for launching @xcite .
\(9 ) we found that jets launched in a setup with smaller diffusive scale height are more perturbed .
the same effect is seen in outflows launched from disks with weaker diffusivity .
\(10 ) we finally remark that it is essential to do long - term simulations covering thousands of rotational periods in order to find a steady state situation of the accretion - ejection dynamics .
our simulations run for 5000 dynamical time steps , corresponding to about 900 revolutions at the inner disk radius as adopted in our reference run . in summary
, we confirm the hypothesis that efficient magneto - centrifugal jet driving requires a strong magnetic flux ( i.e. a low plasma beta ) , together with a large enough magnetic torque in order to produce a powerful jet .
in addition , the magnitude of ( turbulent ) magnetic diffusivity plays the major role in the ejection efficiency , while the anisotropy in the diffusivity mainly affects the jet rotation .
both results imply that the structure of the asymptotic jet is indeed governed by the properties of the accretion disk , here parametrized by the magnetization and magnetic diffusivity
. the mass ejection - to - accretion ratio along with the momentum and energy transfer rates from inflow to outflow are essential properties for any feedback mechanism in star formation or galaxy formation scenarios and could only be derived from simulations resolving the inner region of the jet - launching accretion disk .
we thank the andrea mignone and the pluto team for the possibility to use their code .
we acknowledge helpful and constructive criticism by two referees , which has improved the presentation of the paper . s.s .
acknowledges the warm hospitality by the max planck institute for astronomy .
all simulations were performed on the theo cluster of max planck institute for astronomy .
this work was financed partly by a scholarship of the ministry of science , research , and technology of iran , and by the sfb 881 of the german science foundation dfg .
we normalize all variables , namely @xmath193 to their fiducial values at the inner disk radius .
we have adopted the following typical number values for a yso of @xmath194 and an agn of @xmath195 .
a change in one of the system parameters accordingly changes the scaling of our simulation results which can thus be applied for a variety of jet sources .
the inner disk radius @xmath196 is usually assumed to be a few radii of the central object , @xmath197 where @xmath198 is the schwarzschild radius of the central black hole .
the inner disk radius is usually assumed to be located at the marginally stable orbit at @xmath199 .
since we apply the non - relativistic version of the pluto code , we can not treat any relativistic effects .
we therefore apply a scaling of @xmath200 . for simplicity and convienient comparison with the previous literature
we apply @xmath201au for most comparisons concerning stellar sources .
the orbital velocity at the inner disk radius is @xmath202 for a @xmath201au distance from a yso the orbital speed is @xmath203 .
the mass accretion rate is a parameter which is in principle accessible by observation .
subject to the disk model applied , the observed disk luminosity can be related to an accretion rate . for a yso ,
the accretion rate is typically of the order of @xmath204 , providing a normalisation of the density @xmath205 with @xmath206 applying a length scale @xmath201au and a velocity scale @xmath203 , we obtain @xmath207 normalized value of the magnetic field is obtained by considering the plasma-@xmath104 and the field strength at the equator at the inner radius @xmath208 , @xmath209
an essential point of our setup is to impose an outflow boundary at the outer part of the domain in @xmath74-direction avoiding artificial collimation .
we have implemented a current - free outflow boundary condition which prevents spurious collimation by lorentz forces and which has been thoroughly tested in our previous papers @xcite .
the boundary condition considers a vanishing toroidal electric current density in the ghost cells .
therefore , the lorentz force component perpendicular to the grid boundary vanishes .
the boundary constraints are applied on the poloidal and toroidal component of the electric current density , @xmath210 , @xmath211 , and @xmath212 , respectively . in principle , this is considered for the grid ghost cells @xmath213 , adjunct to the domain boundary at @xmath214 ( see the 3.1.1 in @xcite ) .
the transverse @xmath215 and normal @xmath216 magnetic field components in the domain , together with the transverse field component @xmath217 of the first ghost zone , constitute a toroidal corner - centered electric current @xmath218 . with that a constraint for a current - free @xmath219 boundary condition can been implemented numerically , @xmath220 .
\label{app : outer - boundary}\end{aligned}\ ] ] | we investigate the launching of jets and outflows from magnetically diffusive accretion disks . using the pluto code we solve the time - dependent resistive mhd equations taking into account the disk and jet evolution simultaneously .
the main question we address is _ which kind of disks do launch jets and which kind of disks do not ? _ in particular , we study how the magnitude and distribution of the ( turbulent ) magnetic diffusivity affect mass loading and jet acceleration .
we have applied a turbulent magnetic diffusivity based on @xmath0-prescription , but have also investigate examples where the scale height of diffusivity is larger than that of the disk gas pressure .
we further investigate how the ejection efficiency is governed by the magnetic field strength .
our simulations last for up to 5000 dynamical time scales corresponding to 900 orbital periods of the inner disk . as a general result
we observe a continuous and robust outflow launched from the inner part of the disk , expanding into a collimated jet of super fast magneto - sonic speed . for long time
scales the disk internal dynamics changes , as due to outflow ejection and disk accretion the disk mass decreases . for magneto - centrifugally driven jets
we find that for i ) less diffusive disks , ii ) a stronger magnetic field , iii ) a low poloidal diffusivity , or a iv ) lower numerical diffusivity ( resolution ) , the mass loading of the outflow is increased - resulting in more powerful jets with high mass flux . for weak
magnetization the ( weak ) outflow is driven by the magnetic pressure gradient .
we consider in detail the advection and diffusion of magnetic flux within the disk and we find that the disk and outflow magnetization may substantially change in time .
this may have severe impact on the launching and formation process - an initially highly magnetized disk may evolve into a disk of weak magnetization which can not drive strong outflows .
we further investigate the jet asymptotic velocity and the jet rotational velocity in respect of the different launching scenarios .
we find a lower degree of jet collimation than previous studies , most probably due to our revised outflow boundary condition . |
the watt balance , an experiment proposed by dr b. p. kibble in 1976 @xcite , is widely employed at national metrology institutes ( nmis ) for precisely measuring the planck constant @xmath1 towards the redefinition of one of the si base units , the kilogram @xcite .
the new definition of the kilogram will be realized by fixing the numerical value of the planck constant @xcite , which is expected to be determined with a relative uncertainty of two parts in @xmath2 .
the role of a watt balance experiment is to transfer the mass standard from the only mass in bureau international des poids et mesures ( bipm ) , i.e. , the international prototype of kilogram ( ipk ) , to a value that makes the planck constant exactly equal to @xmath3 j@xmath4s .
the detailed origin , principle and recent progress of the watt balance is presented in several review papers , e.g. , @xcite . here a brief summary of the measurement is given .
the watt balance is operated in two separated measurement modes , conventionally named as the weighing mode and the velocity mode . in the weighing mode ,
the magnetic force produced by a coil with dc current in the magnetic field is balanced by the gravity of a test mass , and a force balance equation can be written as @xmath5 where @xmath6 denotes the magnetic flux density , @xmath7 the current in the coil , @xmath8 the wire length of the coil , @xmath9 the test mass and @xmath10 the local gravitational acceleration . in the velocity mode ,
the coil moves along the vertical direction in the same magnetic field , and generates an induced voltage , i.e. @xmath11 where @xmath12 is the coil velocity and @xmath13 the induced voltage . by combining equations ( [ 1 ] ) and ( [ 2 ] )
, the geometrical factor @xmath14 can be eliminated and a virtual watt balance equation is obtained as @xmath15 in equation ( [ 3 ] ) , the induced voltage @xmath13 is measured by a josephson voltage standard ( jvs ) linked to the josephson effect , i.e. @xmath16 where @xmath17 denotes a known frequency , @xmath18 the electron charge .
the current @xmath7 is measured by the josephson effect in conjunction with the quantum hall effect as @xmath19 where @xmath20 is the voltage drop on a resistor @xmath21 in series with the coil , @xmath22 the known frequency , and @xmath23 an integer number .
a combination of equations ( [ 3])-([5 ] ) yields the expression of the planck constant @xmath1 in the si unit as @xmath24 it has been known that all the quantities on the right side of ( [ 6 ] ) can be measured with a relative uncertainty lower than one part in @xmath25 , and therefore , on the current stage , the planck constant @xmath1 is expected to be determined by watt balances with a relative uncertainty less than @xmath26 for achieving the purpose of redefining the kilogram . the watt balance is considered as one of the most successful experiments to precisely determine the planck constant @xmath1 and is employed in many nmis , e.g. @xcite . in order to generate a kilogram level magnetic force to balance the gravity of a mass while keep the power assumption of the coil as low as possible in the weighing mode , a strong magnetic field , e.g. 0.5 t , is required at the coil position for the watt balance .
accordingly , permanent magnets with high permeability yokes are introduced to achieve the strong magnetic field .
one of such magnetic systems ( shown in figure [ fig2 ] ) , developed by the bipm watt balance group @xcite , is the most preferred . in the shown magnet construction ,
two permanent magnet rings with opposite magnetization poles are installed inside the inner yoke .
the magnetic flux of the permanent magnet rings is guided by high permeability yokes through the air gap .
as the work area of the coil , the air gap , is designed to be long and narrow , a strong magnetic field with good uniformity can be generated along the radial direction @xmath27 . as the total flux through the air gap
is roughly a constant , the magnetic flux density along the radial direction , @xmath28 , decays along @xmath27 direction following an approximative @xmath29 relation . in a @xmath29 magnetic field
, it can be proved that the @xmath14 of the coil is a constant , independent to any coil deformation or horizontal displacements @xcite .
as the magnetic flux is closed with yokes , the shown watt balance magnet has a good feature of self shielding , lowing additional flux exchange between inside and outside of the magnet .
based on these advantages , the shown magnet has been widely adopted by other nmis , such as the federal institute of metrology ( metas ) , switzerland @xcite , the national institute of standards and technology ( nist ) , usa @xcite , the measurement standards laboratory ( msl ) , new zealand @xcite and the korea research institute of standards and science ( kriss ) , south korea @xcite . in reality , since the air gap length is finite and hence the @xmath29 decay of the magnetic field @xmath28 is not true over the whole vertical range , i.e. , the fringe effect will introduce a vertical magnetic component @xmath30 at any coil positions where @xmath31 .
this fact indicates that the unknown magnetic field component can cause alignment problems @xcite , e.g. , the vertical magnetic field component can exert undesired radial force or torque in the weighing mode and could produce additional voltage in the velocity mode . considering a strict requirement of measurement accuracy in watt balances ,
any aspect may bring a systematic error should be carefully analyzed .
an accurate 3d mapping of the magnetic field in the air gap is a useful tool to provide important information on systematic error elimination .
at least three benefits will be created by a full air gap 3d magnetic field mapping .
the first is the misalignment error relaxation : the value of geometrical factor @xmath14 in equations ( [ 1 ] ) and ( [ 2 ] ) can be actually affected by parasitic coil motions when the fringe field is considered .
this misalignment error can be corrected if the 3d magnetic field profile is obtained , which in theory can relax the alignment requirement of a watt balance .
the second benefit is that with knowing a full field profile , a best coil diameter can be chosen in order to employ the maximum flat @xmath0 profile in the air gap , and a flat @xmath0 profile can reduce uncertainties in both velocity and voltage measurements .
thirdly , the damping of the coil is usually applied in the breath between weighing and velocity sweeps . during the breath ,
the damping device is either at the top or the bottom of the air gap , where a coupling between the damping current and both magnetic components ( the radial magnetic field @xmath32 and vertical magnetic field @xmath30 ) should be considered . in this case
, the 3d magnetic field can supply better feedback to the current parameters for simultaneously damping all dimensional motions of the coil .
as is mentioned in @xcite , the narrow air gap makes it difficult to directly measure the global magnetic field profiles .
only profiles of the radial magnetic flux density along the vertical direction , @xmath0 , can be precisely measured by either a high resolution magnetic probe or the gradient coil ( gc ) method @xcite .
a polynomial estimation algorithm based on at least two measurements of @xmath0 profiles has been developed in @xcite to calculate the global magnetic field in the air gap .
the algorithm essentially approximates the magnetic field lines by a polynomial fitting and the approximation precision depends on the fitting order , i.e. the quantities of the measured @xmath0 .
a higher accuracy requires more @xmath0 profile measurements , which in reality is difficult to be done at different radii of the narrow air gap . besides
, the algorithm presented in @xcite would be much more complicated when the estimator order @xmath33 increases , e.g. , @xmath34 .
we later noticed the information in measured @xmath0 is not fully utilized in the polynomial estimation algorithm , and hence in theory it is possible to develop an analytical algorithm based on only one @xmath0 measurement . this motivation leads to an improved analytical algorithm to map the 3d magnetic field in the air gap region , which has been presented in this article .
the new algorithm is based on fundamental electromagnetic natures of the magnet and has advantages in both convenience and accuracy .
the rest of this paper is organized as following . in section [ sec2 ] ,
the new analytical algorithm is presented . in section [ sec3 ] ,
numerical simulations are employed to verify the accuracy of the analytical algorithm .
some discussions on the watt balance magnet design are shown in section [ sec4 ] and a conclusion is drawn in section [ sec5 ] .
a detailed dimension of the air gap in the watt balance magnet is shown in figure [ fig3 ] .
the area @xmath35 , @xmath36 is our model region where @xmath37 and @xmath38 is the inner and outer radii of the air gap and @xmath39 is a half of the air gap height .
the radial magnetic flux density along the vertical direction @xmath40 , or the radial magnetic field along the vertical direction @xmath41 , can be measured by a gc or a magnetic probe at the radial coordinate @xmath42 .
is measured at the horizontal coordinate @xmath42 ( the green line).,scaledwidth=40.0% ] in the analysis , the magnetic scalar potential @xmath43 is chosen as a quantity to be solved .
the magnetic field @xmath44 can be expressed as the negative gradient of the magnetic scalar potential , i.e. , @xmath45 .
@xmath44 is the sum of two components in an axisymmetrical coordinate system as @xmath46 where @xmath47 and @xmath48 denote the unit vectors in the @xmath27 and @xmath49 directions ; @xmath50 and @xmath51 are two components of the magnetic field in @xmath27 and @xmath49 directions . as @xmath45 , equation ( [ eq1 ] ) can also be written as @xmath52 the magnetic scalar potential @xmath43 in the air gap area , i.e. @xmath53 ( @xmath37 , @xmath38 ) , @xmath54 ( @xmath55 , @xmath39 ) , can be described by the laplace s equation as @xmath56 the separation of variables is applied to solve equation ( [ eq2 ] ) . in this case , the magnetic scalar potential @xmath43 is supposed to be expressed as the product of two independent functions @xmath57 and @xmath58 , where @xmath57 and @xmath58 are functions of a single variable , i.e. @xmath27 and @xmath49 . substituting @xmath59 into equation ( [ eq2 ] ) yields @xmath60 since
the two terms on the left side of equation ( [ sep_vars ] ) are independent in dimension , equation ( [ sep_vars ] ) can be written as @xmath61 where @xmath62 is a constant .
note that the solution form of equation ( [ sep_vars ] ) will be determined by the sign of @xmath62 .
when @xmath63 , the solution can be written as @xmath64 where @xmath65 , @xmath6 , @xmath66 and @xmath67 are constants .
when @xmath68 ( e.g. @xmath69 ) , the solution can be obtained as @xmath70 \left[{{c_1}{j_0}(\lambda r ) + { d_1}{y_0}(\lambda r ) } \right ] , \end{array}\ ] ] where @xmath71 and @xmath72 denote hyperbolic cosine function and hyperbolic sine function ; @xmath73 and @xmath74 are the zeroth order bessel functions of the first and second kinds , @xmath75 , @xmath76 , @xmath77 , @xmath78 and @xmath79 are all constants . when @xmath80 ( e.g. @xmath81 ) , the solution can be written as @xmath82 \left [ { { c_2}{i_0}(\rho r ) + { d_2}{k_0}(\rho r ) } \right ] , \end{array}\ ] ] where @xmath83 and @xmath84 denote the zeroth order modified bessel functions of the first and second kinds , @xmath85 , @xmath86 , @xmath87 , @xmath88 and @xmath89 are all constants . the general solution of equation ( [ sep_vars ] ) is the linear combination of @xmath90 , @xmath91 and @xmath92 , i.e. @xmath93 \left[c_{1n}j_0(\lambda_n r)+d_{1n}y_0(\lambda_n r)\right]\\ + \sum\limits_{n=1}^{\infty}\left[a_{2n}\cos ( \rho_n z)+b_{2n}\sin ( \rho_n z)\right ] \left[c_{2n}i_0(\rho_n r)+d_{2n}k_0(\rho_n r)\right ] .
\label{eq.general } \end{array}\end{aligned}\ ] ] based on equations ( [ eq1.1 ] ) and ( [ eq.general ] ) , the vertical magnetic field component @xmath51 can be expressed as @xmath94 \left[c_{1n}j_0(\lambda_n r)+d_{1n}y_0(\lambda_n r)\right]\\ + \sum\limits_{n=1}^{\infty}\rho_n\left[a_{2n}\sin ( \rho_n z)-b_{2n}\cos ( \rho_n z)\right ] \left[c_{2n}i_0(\rho_n r)+d_{2n}k_0(\rho_n r)\right ] .
\label{eq4 } \end{array}\end{aligned}\ ] ] since the yoke permeability in the presented watt balance magnet is very high , two yoke - air boundaries , i.e. @xmath95 and @xmath96 , can be considered as equipotential surfaces . as a result
, the vertical magnetic field along the vertical direction should be zero at both @xmath95 and @xmath96 , i.e. , @xmath97 . using this boundary condition ,
the following equations are established , i.e. @xmath98 @xmath99 and @xmath100 in equation ( [ eq.condition11 ] ) , since @xmath101 is a function of monotonicity where @xmath102 , we have @xmath103 when @xmath104 , and hence @xmath105 . in equation ( [ eq.condition12 ] )
, we can set @xmath106 and @xmath107 to ensure the condition , i.e. , @xmath108 , to be always satisfied . in this case , @xmath109 should be set to values to establish another condition , i.e. @xmath110 . in equation ( [ eq.condition13 ] ) , because @xmath111 is a monotone increasing function while @xmath112 is a monotone decreasing function , we have @xmath113 and @xmath114 .
therefore , @xmath115 and @xmath116 is obtained .
the distribution of @xmath43 is symmetrical about the line @xmath121 and thus the odd symmetrical function @xmath122 should be removed from the expression of @xmath43
. then equation ( [ eq.after_condition3 ] ) is reduced to @xmath123 .
\label{eq.after_condition2 } \end{array}\ ] ] using the measured profile @xmath124 , the remaining unknown constants in equation ( [ eq.after_condition2 ] ) can be solved .
based on equations ( [ eq1 ] ) and ( [ eq.after_condition2 ] ) , @xmath124 can be written as @xmath125\\ \displaystyle = c'+\sum\limits_{n=1}^{\infty}a'_n \cosh(\lambda_n z ) , \label{eq.compare1 } \end{array}\end{aligned}\ ] ] where @xmath126 and @xmath127 denote the first order bessel functions of the first and second kinds ; @xmath128 ; @xmath129 $ ] ( @xmath130 ) .
for convenience of expression , we can set @xmath131 and @xmath132 , and then @xmath133 . as a result , equation ( [ eq.compare1 ] ) can be expressed as @xmath134 the values of @xmath135 can be obtained by expanding @xmath41 in forms of @xmath136 .
however , the implementation process is complicated because of the non - orthogonality of the base functions @xmath136 , i.e. @xmath137 where @xmath138 is defined as the inner product of two functions @xmath139 and @xmath140 . to reduce the complexity due to the non - orthogonality , in this paper
, the process of obtaining @xmath135 is as follows : first , the base functions @xmath136 are transformed into orthogonal normalized base functions @xmath141 , then @xmath41 is expanded in forms of @xmath141 in a convenient way , and finally base functions @xmath141 are replaced with @xmath136 . in this way , @xmath41 can be easily expanded in forms of @xmath136 and hence @xmath135 is obtained . to practise the above idea , here
the gram - schmidt orthogonalization procedure is employed to transform @xmath136 into orthogonal normalized base functions @xmath141 .
the definition of the norm of function @xmath139 is expressed as @xmath142 then base functions @xmath136 are orthogonalized and normalized through a recursion formula , expressed as @xmath143 in recursion formula ( [ eq.orthogonal_2 ] ) the inner product @xmath144 when @xmath145 and @xmath146 when @xmath147 .
after the orthogonalization and normalization procedure , @xmath41 can be expanded in forms of @xmath141 , i.e. @xmath148 where the coefficient @xmath149 can be calculated by solving the inner product of @xmath41 and @xmath141 , because @xmath150 the last step of obtaining @xmath135 is replacing @xmath141 in equation ( [ eq.orthogonal_3 ] ) with @xmath151 , and thus @xmath141 should be solved by expressing it with combinations of @xmath136 .
the relation between @xmath136 and @xmath141 has been shown in recursion formula ( [ eq.orthogonal_2 ] ) .
it is found difficult to directly express @xmath141 as a linear combination of @xmath136 , but instead it is easy to write @xmath136 as a linear combination of @xmath141 , i.e. @xmath152 where @xmath153 , @xmath154 and @xmath155 are constants that have been solved in equation ( [ eq.orthogonal_2 ] ) . setting @xmath156 and @xmath157 as column vectors as @xmath158 equation ( [ eq.orthogonal_6 ] ) can be written as the product of a matrix and the column vector @xmath159 , i.e. @xmath160 where @xmath161 denotes a lower triangular matrix , i.e. @xmath162 based on equation ( [ matrix1 ] ) , @xmath141 can be solved as the linear combination of @xmath136 , i.e. @xmath163 where @xmath164 denotes the element on line numbered @xmath165 column numbered @xmath166 of @xmath167 , the inverse of the matrix @xmath168 . knowing the @xmath141 expression with a linear combination of @xmath136 , the unknown constants @xmath135 in equation ( [ eq.compare11 ] ) can be solved .
replacing @xmath141 in equation ( [ eq.orthogonal_3 ] ) with equation ( [ ex ] ) , @xmath41 is written as the combination of @xmath136 , i.e. @xmath169\\ = \sum\limits_{i=0}^{\infty}\big[\sum\limits_{n=0}^{\infty}f_n\mathbf{m}^{-1}(n , i)\big]\cosh(\lambda_i z)\\ = \sum\limits_{n=0}^{\infty}\big[\sum\limits_{i=0}^{\infty}f_i\mathbf{m}^{-1}(i , n)\big]\cosh(\lambda_n z ) , \label{eq.compare2 } \end{array}\end{aligned}\ ] ] comparing equations ( [ eq.compare11 ] ) and ( [ eq.compare2 ] ) , @xmath135 is solved as @xmath170 accordingly , the constant @xmath66 in equation ( [ eq.compare1 ] ) is calculated as @xmath171 , \label{eq.c } \end{array}\end{aligned}\ ] ] and @xmath172 in equation ( [ eq.compare1 ] ) is solved as @xmath173}\\ \displaystyle=\frac{\sum\limits_{i=0}^{\infty}f_i\mathbf{m}^{-1}(i , n)}{\lambda_n\left[j_1(\lambda_n r_0)y_0(\lambda_n b)-j_0(\lambda_n b)y_1(\lambda_n r_0)\right]}. \label{eq.a1n } \end{array}\end{aligned}\ ] ] by a combination of equations ( [ eq.after_condition2 ] ) , ( [ eq.c ] ) and ( [ eq.a1n ] ) , the magnetic scalar potential @xmath43 is obtained as @xmath174(\ln r-\ln b)\\ + \sum\limits_{n=1}^{\infty}\left[\sum\limits_{i=0}^{\infty}f_i\mathbf{m}^{-1}(i , n)\right ] \displaystyle\frac{\cosh ( \lambda_n z)[j_0(\lambda_n r)y_0(\lambda_n b)-j_0(\lambda_n b)y_0(\lambda_n r)]}{\lambda_n\left[j_1(\lambda_n r_0)y_0(\lambda_n b)-j_0(\lambda_n b)y_1(\lambda_n r_0)\right]}. \label{eq.solution } \end{array}\end{aligned}\ ] ] based on equation ( [ eq1 ] ) , the two magnetic field components @xmath51 and @xmath50 can be calculated respectively as @xmath175 \displaystyle\frac{\sinh ( \lambda_n z)\left[j_0(\lambda_n r)y_0(\lambda_n b)-j_0(\lambda_n b)y_0(\lambda_n r)\right]}{j_1(\lambda_n r_0)y_0(\lambda_n b)-j_0(\lambda_n b)y_1(\lambda_n r_0 ) } , \label{eq.hz1 } \end{array}\ ] ] and @xmath176 \displaystyle\frac{\cosh ( \lambda_n z)\left[j_1(\lambda_n r)y_0(\lambda_n b)-j_0(\lambda_n b)y_1(\lambda_n r)\right]}{j_1(\lambda_n r_0)y_0(\lambda_n b)-j_0(\lambda_n b)y_1(\lambda_n r_0)}. \label{eq.hr1 } \end{array}\end{aligned}\ ] ] it can be proved that the magnetic field solutions obtained in equations ( [ eq.hz1 ] ) and ( [ eq.hr1 ] ) are unique . the proof of the uniqueness theorem is attached in the appendix .
over @xmath79 ( the blue curve ) and the horizontal axis ( the red curve).,scaledwidth=55.0% ] in order to evaluate the analytical algorithm accuracy , numerical simulations based on the finite element method ( fem ) are performed . in these fem simulations ,
the parameters are set close to a 1:1 real watt balance magnet : @xmath37 , @xmath38 , @xmath39 and @xmath42 are set as @xmath177 mm , @xmath178 mm , @xmath179 mm and @xmath180 mm respectively ; the relative permeability of the yoke is set as @xmath181 and the magnetic strength of the permanent magnet is set as 800kam@xmath182 in the vertical direction . in the simulation , the magnetic flux density profile @xmath32(210 mm , @xmath49 ) is calculated by fem simulation as a known condition of the presented analytical algorithm to simulate the actual measurements of either a gc coil or a magnetic probe .
note that in reality it is impossible to calculate infinite number of terms @xmath165 in equation ( [ eq.solution ] ) , and hence the first nine @xmath109 are adopted .
the solution of the magnetic scalar potential @xmath43 , accordingly , can be written as @xmath183(\ln r-\ln b)\\ + \sum\limits_{n=1}^{9}\left[\sum\limits_{i=0}^{9}f_i\mathbf{m}^{-1}(i , n)\right ] \displaystyle\frac{\cosh ( \lambda_n z)[j_0(\lambda_n r)y_0(\lambda_n b)-j_0(\lambda_n b)y_0(\lambda_n r)]}{\lambda_n\left[j_1(\lambda_n r_0)y_0(\lambda_n b)-j_0(\lambda_n b)y_1(\lambda_n r_0)\right]}. \label{solution_4 } \end{array}\end{aligned}\ ] ] in equation ( [ solution_4 ] ) , @xmath184 , as demonstrated in solving equation ( [ eq.condition12 ] ) , should satisfy @xmath110 , and their values are calculated by a numerical method . as shown in figure [ figure2 ] , the solutions @xmath109 are the intersections of function @xmath185 and the horizontal axis .
the first nine intersections are solved as @xmath186mm@xmath182 ( @xmath187 ) .
as presented in equation ( [ eq.orthogonal_2 ] ) , the base functions @xmath136 are transformed into orthogonal normalized functions @xmath188 , where @xmath189 is set to @xmath190 , as it is defined in section [ sec2 ] .
using equation ( [ eq.orthogonal_31 ] ) , constants @xmath191 are solved as @xmath192 the @xmath161 in this example is a @xmath193 matrix , whose elements have been calculated based on equation ( [ eq.matrix ] ) , i.e. @xmath194 the matrix @xmath195 is obtained by inversing @xmath161 , and two magnetic field components , i.e. , @xmath51 and @xmath50 , then can be solved based on equations ( [ eq.hz1 ] ) and ( [ eq.hr1 ] ) . figure [ figure25 ] shows the comparison of @xmath196 results calculated by the fem simulation and the analytical equation in ( [ eq.hr1 ] ) , where @xmath197 and @xmath198 denotes the vacuum ( air ) permeability .
it can be seen that the two curves agree well with each other , which indicates that the order of @xmath165 is high enough to estimate @xmath40 . in order to further check the agreement with different numbers of @xmath165 , a relative fit error @xmath199 is defined , i.e. @xmath200 where @xmath201 is the measurement magnetic flux density , @xmath202
the fit value of the magnetic flux density , and @xmath203 the average of @xmath202 , is analyzed on its decay rate over @xmath165 .
the calculation result is shown in figure [ figure26 ] .
it can be seen that the fit accuracy is mainly limited by the magnetic field profile measurement . when @xmath204 , the fit residual is already comparable to the sensitivity of the magnetic field measurement instruments , i.e. the accuracy of fem simulation in this case .
obtained by fem simulation ( the blue curve ) and the analytical method ( the red curve).,scaledwidth=55.0% ] over different orders @xmath165.,scaledwidth=55.0% ] figure [ figure4 ] shows the calculation result of the 3d magnetic flux density @xmath205 and @xmath206 based on the analytical equations ( [ eq.hz1 ] ) and ( [ eq.hr1 ] ) . the air gap region @xmath53(200 mm , 230 mm ) , @xmath54(-50 mm , 50 mm ) , where the watt balance is conventionally operated , has been focused .
the calculation result clearly shows the fringe effect : the absolute value of the vertical magnetic flux density @xmath207 increases at both vertical ends of the air gap ; the horizontal magnetic flux density component @xmath208 is also bent from the @xmath29 decay surface . to evaluate the accuracy of the presented analytical algorithm ,
some typical @xmath209 and @xmath196 curves given by fem simulation and the analytical equations ( [ eq.hz1 ] ) and ( [ eq.hr1 ] ) are compared in figure [ figure3 ] .
it can be seen that the calculation obtained by the analytical algorithm agrees very well with that simulated by fem . for a full view of the field difference of fem and the analytical algorithm ,
the differential maps of @xmath207 and @xmath208 are calculated in the air gap region @xmath53(200 mm , 230 mm ) , @xmath54(-50 mm , 50 mm ) and the calculation results are shown in figure [ figure5 ] . .
( b ) magnetic flux density distribution of @xmath208 .
, scaledwidth=90.0% ] and @xmath196 curves obtained by the analytical method and fem .
the blue curves are obtained by fem simulation and the red curves are calculated by the analytical method .
( a ) the @xmath209 curves , which from top to bottom are presented with different values of @xmath210 , i.e. 50 mm , 45 mm , 40 mm , 30 mm , 0 mm , -30 mm , -40 mm , -45 mm and -50 mm .
( b ) the @xmath196 curves , which from top to bottom have different values of @xmath42 , i.e. 201 mm , 205 mm , 210 mm , 215 mm , 220 mm , 225 mm and 229mm.,scaledwidth=90.0% ] for comparison of the two analytical algorithms , the field difference between the fem simulation and the polynomial estimation algorithm in @xcite has been also calculated with the same permeability . to quantize the comparison ,
the average difference between the analytical method and the fem is defined as @xmath211 where @xmath212 is the radical index number and @xmath213 is the vertical index number ; @xmath214 is the magnetic flux density difference between the proposed analytical method and fem simulation .
the polynomial estimation algorithm is adopted in its highest precision case , i.e. three different @xmath0 profiles are measured in radii @xmath215 mm , @xmath216 mm and @xmath180 mm .
the average differences of the polynomial estimation algorithm are 0.0595mt for @xmath207 and 0.0894mt for @xmath208 while the average difference of the new analytical algorithm is 0.0060mt for @xmath207 and 0.0048mt for @xmath208 .
the accuracy of the new analytical algorithm in this case is respectively 10 and 20 times better for @xmath207 and @xmath208 than that of the polynomial estimation algorithm . ;
( b ) the magnetic flux density error of @xmath206 .
the bottom subgraphs are the calculation error between the polynomial estimation algorithm and fem simulation : ( c ) magnetic flux density error of @xmath205 ; ( d ) the magnetic flux density error of @xmath206.,scaledwidth=95.0% ]
the numerical simulation in section [ sec3 ] exhibits advantages in field representation accuracy and convenience in measurement for the new analytical algorithm .
it should be emphasized that the shown analytical algorithm , as well as the polynomial estimation algorithm proposed in @xcite , is based on the assumption that the normal component of the magnetic field on the air - yoke boundary strictly equals zero .
however , in reality a weak normal component exists on the air - yoke boundaries duo to the finite permeability of the yoke , which creates the main part of the calculation error . for a further check , we calculated the magnetic field difference with different values of the yoke permeability .
figure [ fig9 ] shows the calculation errors between the proposed analytical algorithm and fem simulation when @xmath217 and @xmath218 .
it can be seen that the difference error is getting smaller when the permeability is higher .
this indicates that the analytical algorithm will have better performance when applied in the watt balance magnet with high permeability , e.g. the bipm watt balance magnet . ;
( b ) the calculation error of @xmath206 .
the bottom subgraphs are the calculation error map when the relative permeability is set as @xmath219 : ( c ) the calculation error of @xmath205 ; ( d ) the calculation error of @xmath206.,scaledwidth=95.0% ] .
( b ) the simulation results of @xmath220mm@xmath221 and @xmath222mm@xmath221 .
@xmath223 is defined as @xmath224 where @xmath225 is the average value of @xmath0.,scaledwidth=95.0% ] as mentioned in section [ sec1 ] , one of the main purpose for presenting the 3d magnetic field in the air gap is to find a best coil diameter for obtaining a widest @xmath0 profile . here
we give a suggestion by taking the calculation in section [ sec3 ] as an example .
knowing the 3d magnetic field , the coordinate @xmath226 where @xmath227 has a peak value , i.e. @xmath228\\ \displaystyle\times\frac{\lambda_n\sinh ( \lambda_n z_0)\left[j_1(\lambda_n r_{\rm p})y_0(\lambda_n b)-j_0(\lambda_n b)y_1(\lambda_n r_{\rm p})\right]}{j_1(\lambda_n r_0)y_0(\lambda_n b)-j_0(\lambda_n b)y_1(\lambda_n r_0)}=0 , \label{new } \end{array}\ ] ] can be solved .
the calculation result of @xmath226 has been shown in figure [ figure10](a ) .
it shows that for the typical watt balance magnet , a best coil radius should be smaller than the air gap center radius .
if the velocity sweep range is set @xmath229 in the velocity mode , the best coil radius is calculated by means of the @xmath230 as @xmath231 for the example shown in figure [ figure10](a ) , @xmath232 mm , @xmath233 mm , and @xmath234 is calculated as 214.54 mm . by the calculation , the @xmath220mm@xmath221 should have a wider flat profile than @xmath222mm@xmath221 . to check it
, we have plotted these two @xmath0 profiled in figure [ figure10](b ) by fem and the result verifies the conclusion .
an analytical algorithm , employing only one measured @xmath0 , is presented for calculating the 3d magnetic field profile in watt balance magnets . compared to the polynomial estimation algorithm @xcite
, the new algorithm is based on fundamental electromagnetic natures of the magnet and has significant advantages in both convenience and accuracy .
it is shown that the new analytical algorithm can improve the field mapping accuracy by more than one magnitude than that of a simple polynomial estimation , which has a good potential in application of high permeability cases , e.g. , the bipm magnet .
the presented work can supply necessary information for misalignment analysis and parameter determinations in watt balances .
base on the study , a best coil radius , which should be designed slightly smaller than the air gap center , is suggested .
the discussion shows that the accuracy of the proposed analytical algorithm is mainly limited by a finite permeability of the yoke material .
therefore , a correction model of yoke permeability should be focused in a following investigation .
also , in this paper the @xmath27 symmetry of the watt balance magnet is assumed , which in reality may be not true .
further studies with considerations of the @xmath27 asymmetry of the magnet system may be addressed in the future .
the authors would like to thank mr .
xuanyu dong for advice about using the gram - schmidt orthogonalization procedure and dr qing wang at durham university , uk for language proofing .
this work is supported by the national natural science foundation of china ( grant no .
51507088 ) .
in this appendix , the uniqueness of magnetic field solutions , i.e. , @xmath50 and @xmath51 in equations ( [ eq.hz1 ] ) and ( [ eq.hr1 ] ) , is proved by a reductio ad absurdum .
we assume that the solution of equation ( [ eq.solution ] ) is not unique in the magnet air gap region @xmath235 , @xmath236 . without losing generality , two different solutions @xmath237 and @xmath238
are supposed to satisfy the condition @xmath239 , @xmath240 and @xmath41 .
based on the symmetry of the air gap , the magnetic scalar potential at @xmath241 for @xmath237 and @xmath238 , i.e. @xmath242 , @xmath243 , @xmath244 and @xmath245 , should meet @xmath246 according to the uniqueness theorem for static magnetic fields , @xmath239 , @xmath240 , @xmath242 and @xmath243 form the boundary conditions of region @xmath235 , @xmath236 and determines a unique solution of @xmath237 , which is expressed similarly as equation ( [ eq.after_condition2 ] ) , i.e. @xmath247 .
\end{array } \label{eq.app1}\end{aligned}\ ] ] the unknown constants in equation ( [ eq.app1 ] ) , i.e. @xmath66 and @xmath172 , can be solved by expanding @xmath242 in forms of fundamental functions @xmath248 and @xmath249 .
the solved constants are symbolized as @xmath250 and @xmath251 , and then the solution @xmath237 can be written as @xmath252 .
\label{eq.phim1 } \end{array}\end{aligned}\ ] ] similarly , @xmath238 is the unique solution with boundary conditions @xmath239 , @xmath240 , @xmath244 and @xmath245 , which can be expressed as @xmath253 .
\label{eq.phim2 } \end{array}\end{aligned}\ ] ] based on equations ( [ eq1 ] ) , ( [ eq1.1 ] ) and ( [ eq.phim1 ] ) , the radial component of the magnetic field at @xmath42 , i.e. @xmath254 , can be expressed as @xmath255\cosh ( \lambda_n z),~~~~ \end{array}\end{aligned}\ ] ] and based on equations ( [ eq1 ] ) , ( [ eq1.1 ] ) and ( [ eq.phim2 ] ) , the expression of @xmath256 is obtained as @xmath257\cosh ( \lambda_n z).~~~~ \end{array}\end{aligned}\ ] ] it has been supposed that both @xmath237 and @xmath238 can both satisfy the condition @xmath41 , therefore @xmath258 . as is known , a function of @xmath259 can not be expressed by linear combinations of @xmath136 where @xmath260 , thus @xmath261 and @xmath262 are established .
further , based on equations ( [ eq.phim1 ] ) and ( [ eq.phim2 ] ) , we have @xmath263 obviously , equation ( [ eq.appxz ] ) and the stated assumption @xmath264 are contradictory , and hence the supposition is false and the theorem , i.e. the solution of equation ( [ eq.solution ] ) is unique , is valid .
therefore @xmath50 and @xmath51 in equations ( [ eq.hz1 ] ) and ( [ eq.hr1 ] ) are the unique solutions .
et al _ 2015 field representation of a watt balance magnet by partial profile measurements _ metrologia _ 445 - 453 kibble b p 1976 a measurement of the gyromagnetic ratio of the proton by the strong field method _ atomic masses and fundamental constants 5 _ pp 545 - 551 | a yoke - based permanent magnet , which has been employed in many watt balances at national metrology institutes , is supposed to generate strong and uniform magnetic field in an air gap in the radial direction .
however , in reality the fringe effect due to the finite height of the air gap will introduce an undesired vertical magnetic component to the air gap , which should either be measured or modeled towards some optimizations of the watt balance . a recent publication , i.e. , _ metrologia _ 52(4 ) 445 @xcite , presented a full field mapping method , which in theory will supply useful information for profile characterization and misalignment analysis .
this article is an additional material of @xcite , which develops a different analytical algorithm to represent the 3d magnetic field of a watt balance magnet based on only one measurement for the radial magnetic flux density along the vertical direction , @xmath0 .
the new algorithm is based on the electromagnetic nature of the magnet , which has a much better accuracy . |
the descriptions of the total cross sections for elastic hadron - hadron , photon - hadron , and lepton - hadron scattering at large energy have been important issues in the understanding of qcd . only a limited type of hadron reactions , the so - called hard processes , could be understood within the perturbative qcd .
most of the available data , however , deals with processes where the momentum transfer between the quarks and the gluons is relatively small and therefore their understanding should be the subject of nonperturbative studies .
the latter are performed by the use of effective theories based on qcd .
regge theory is a well - known nonperturbative approach to hadron reactions @xcite .
it is based on the assumption that at large energy multi - particle exchanges with definite quantum numbers can be expressed by one effective particle exchange with its propagator given by the so - called regge trajectory @xmath6 .
it was shown that one can fix the shape of the usual regge trajectory @xmath7 from the analysis of the mass spectrum @xmath8 of the hadronic states with spin @xmath9 , where @xmath10 lies on this trajectory .
most regge trajectories have the same slope @xmath11 gev@xmath12 , which can be related with the universal value of the string tension between quarks , as has been found in lattice qcd calculations
. the intercepts of the usual secondary regge trajectories are not very large : @xmath13 .
the pomeron trajectory , which was introduced into particle physics more than forty years ago , has properties which are very different from those of the other trajectories : it has the intercept larger than one , @xmath14 , and a slope @xmath15 gev@xmath12 .
these peculiar properties of the pomeron play an important role in particle reactions , e.g. , the pomeron - exchange not only gives the main contribution to the total hadronic cross sections but also determines the behavior of the elastic differential cross sections at large energy and small momentum transfer .
the origin of the pomeron - exchange has not yet been understood from qcd , but general wisdom ascribes its existence to the the _ conformal anomaly _ of the theory .
this anomaly leads to finite hadron masses due to the non - vanishing of the matrix elements @xmath16 and vacuum energy density .
the landshoff - nachtmann model @xcite directly relates the properties of the soft pomeron with the non - zero value of gluon condensate @xmath17 and the nonperturbative two - gluon exchange between hadrons .
in addition to the conformal anomaly , qcd has also a non - conserved flavor singlet axial vector current due to the _ axial anomaly_. this anomaly has played a crucial role in understanding the pseudoscalar meson spectrum .
however only recently it has been clarified its importance for understanding the dis spin - dependent structure functions . a solution to the proton spin problem ( for a review see , e.g. , ref .
@xcite ) due to the non - vanishing of the matrix element @xmath18 has been suggested , where @xmath19 is the dual to the gluon tensor @xmath20 . within supersymmetric qcd
only some definite chiral combinations of these two tensors , @xmath21 , correspond to superfields @xcite .
therefore the matrix elements @xmath22 and @xmath23 can be related to each other .
this link was used in ref .
@xcite to estimate the value of the flavor singlet axial vector charge of the proton .
does this relation hold also in real qcd ? , and if so , what are its implications in regge phenomenology ?
if this relation holds , it would be natural that , _
in addition to _ the pomeron trajectory , a new anomalous regge trajectory would exist , associated to the vacuum properties of qcd through the axial anomaly and therefore with its quantum numbers .
our aim here is to show through data analysis that this possibility is indeed realized .
we will estimate the contribution of this trajectory to the elastic proton - proton scattering cross section and show it to be very important .
we also compute the contribution of this trajectory to the photoproduction of the @xmath24 and @xmath3 mesons .
moreover we show that due to the specific quantum numbers of the possible exchanges , polarized and unpolarized proton - proton scattering and photoproduction of vector mesons can be very useful for investigating the possible consequences of this new trajectory .
the remaining part of this paper is organized as follows . in sec .
ii the connection of the properties of the @xmath25 meson with axial anomaly is discussed .
the effect of the new trajectory in proton - proton elastic scattering is evaluated in sec .
section iv is devoted to the contribution of the @xmath25-exchange to the cross sections in @xmath24 and @xmath3 meson photoproduction . in sec .
v the contribution of the new trajectory to the polarization observables in @xmath26 collision and vector - meson photoproduction is obtained and its nontrivial role in polarized particle reactions is discussed .
we give a short summary and some conclusions in sec .
the small value of the flavor singlet axial vector charge @xmath28 which was measured by emc @xcite , led to the crisis of the naive parton model for the polarized deep - inelastic scattering @xcite . despite the understanding that the fundamental origin of this phenomenon is related to the non - conservation of the flavor singlet axial vector current due to axial anomaly , the explicit calculation for the value of the flavor singlet axial vector charge within qcd is absent so far @xcite .
the main problem here is the poorly known nonperturbative sector of qcd , which does not allow to perform exact calculations .
one possible way to explain the small value of @xmath28 is to investigate its connection with the effects of the axial anomaly in meson spectroscopy , and there have been many trials to connect the polarized dis anomaly with the anomalous properties of the @xmath29 meson , e.g. through the u(1)@xmath30 problem .
however , it should be mentioned that because of its quantum numbers the pseudoscalar @xmath29 meson can not give a _ direct _ contribution to the double spin asymmetry in polarized dis , which was used to extract the value of the forward flavor singlet _ axial vector _ current matrix element , @xmath31 where @xmath32 is the proton mass and @xmath33 is its spin .
it is evident that only flavor singlet axial vector mesons can contribute to the left hand side of this equation .
there are three axial vector mesons with the appropriate quantum numbers [ @xmath34 contribute to this matrix element : @xmath35 , @xmath36 , and @xmath37 @xcite .
the anomalous value of the nucleon matrix element of the singlet axial vector current implies an anomalous mixing for these axial vector mesons , similar to what happens in the case of the pseudoscalars .
this mixing implies a strong violation of the ozi rule and therefore a strong gluonic admixture @xcite . from the point of view of the present investigation
the mixing is not important , what really matters is the role played by the lightest meson in this channel , the @xmath35 , in the whole procedure .
this meson behaves in the vector axial three isosinglet case in the same way as the @xmath38 behaves in the pseudoscalar case , namely it saturates the anomaly .
it is the purpose of this paper to estimate the effects of the @xmath35-exchange in elastic nucleon - nucleon scattering and vector - meson photoproduction , and in so doing display the special role played by its regge trajectory , as a consequence of the properties of the vacuum of qcd . throughout this paper
, @xmath25 stands for the @xmath35 state and we will neglect the possible admixture of the other hadronic states to the wave function of the @xmath25-meson .
let us briefly review some of the results of sec .
iii of ref .
@xcite to adapt them to our purpose .
the matrix element of the axial vector current can be rewritten as a sum over all possible intermediate states by using the idea of the axial vector dominance : @xmath39 where @xmath40 s are the axial vector meson states characterized by their corresponding flavor quantum numbers .
this expression provides relations between the axial charges of the nucleon and the coupling constants of the axial vector mesons with the nucleon , @xmath41 given by @xmath42 here the decay constants are defined by @xmath43 and @xmath44 is the polarization vector of the meson .
the eighth component of the flavor octet @xmath45 will be identified with the @xmath36 @xcite . from the data on the decay of @xmath46 one obtains @xcite
@xmath47 then the first equation in eq .
( [ gnn ] ) gives @xmath48 the analogue with the @xmath49 and @xmath50 reduced matrix elements for the axial vector currents allows to estimate the coupling constant of the flavor singlet axial vector meson with the nucleon as in ref .
@xcite : @xmath51 note if we use the su(3 ) relations for the decay constants of the axial vector meson octet , @xmath52 , and the su(6 ) relation between @xmath53 and @xmath54 , @xmath55 , then we obtain a larger value , @xmath56 .
the last equation in eq . ( [ gnn ] ) gives @xmath57 by using the axial vector charge @xmath58 which was found recently @xcite by fitting the world data on the spin - dependent structure function @xmath59 .
it should be mentioned that the estimates ( [ singlet ] ) and ( [ sidor ] ) are probably just _ upper _ limits for the corresponding magnitudes , because of the uncertainties in the extrapolation of flavor singlet part of the spin dependent structure function @xmath59 to the low @xmath60 region . at any rate
we find that the violation of the su(6 ) symmetry for the decay constants of axial vector mesons is very large and @xmath61 .
a similar effect was discussed by veneziano and shore @xcite for the case of pseudoscalar mesons .
they have shown that the small value of the decay constant of the flavor singlet @xmath62 meson and @xmath28 is ascribed not to the properties of this meson itself but to the general properties of the qcd vacuum , namely to the phenomena of topological charge screening in the vacuum , @xmath63 where @xmath64 and @xmath65 is the so - called topological susceptibility , @xmath66 then the smallness of the decay constant @xmath67 can be thought to have the same origin , since using eqs .
( [ gnn ] ) and ( [ scr ] ) one can relate this constant with @xmath68 , @xmath69 where we have neglected the mass difference of the @xmath25 and @xmath45 . from all of the above discussion
we conclude that the properties of the @xmath25 are very different from those of the other mesons that belong to the same su(6 ) multiplet , e.g. the @xmath70 and @xmath36 mesons .
furthermore they imply a dominating behavior in the cross sections when the @xmath25 meson contributes and which leads in general to unexpected and thereby anomalous results . in order to establish these facts in a concrete manner
we proceed to calculate the contribution of the @xmath25 meson for two processes , elastic @xmath26 scattering at large @xmath71 and elastic photoproduction of @xmath2 and @xmath3 mesons .
these calculations will indeed show that the exchange of the @xmath25 meson produces important contributions in these processes .
before we conclude this section we would like to compare the @xmath25 with the anomalous pomeron , for the reason that have been mentioned in the introduction .
equation ( [ scr ] ) follows from the consideration of the _ axial anomaly _
contribution to the non - conservation of the flavor singlet axial vector current , @xmath72 where @xmath73 is the number of flavors and @xmath74 is the topological charge density . due to its relation with the axial anomaly ,
the properties of the @xmath25-exchange are related with the distribution of the topological charge in qcd vacuum .
it is well - known that nonperturbative fluctuations of gluon fields called instantons give rise to the nontrivial topological structure of the qcd vacuum ( see a recent review in ref .
the average size of the instantons in this vacuum is much smaller than the confinement radius , @xmath75 therefore the @xmath25-exchange should contribute to the hadronic cross sections at higher values of the momentum transfer than the pomeron - exchange whose @xmath76-dependence is determined approximately by the isoscalar electromagnetic form factor of the hadrons participating in the process @xcite . for the @xmath77 vertex , we shall use the flavor singlet axial vector form factor @xmath78 with @xmath79 gev .
this is comparable to the estimate of the isosinglet axial form factor of ref .
@xcite , which gives @xmath80 using the skyrme model .
recent experimental data @xcite on the flavor octet axial form factor favor @xmath81 .
this form factor decreases slower with increasing @xmath71 than the isoscalar electromagnetic form factor of the pomeron - nucleon vertex @xcite : @xmath82 where @xmath83 is the usual dipole fit .
therefore the @xmath25 contribution will dominate over the pomeron contribution at large @xmath71 . in this paper
we will show explicitly this behavior in elastic proton - proton scattering and vector - meson photoproduction .
one very interesting feature of the available data on hadron - hadron scattering at high energy is their universality @xcite .
this means that at low momentum transfer @xmath84 gev@xmath85 and large energy @xmath86 gev all data are described rather well by the pomeron - exchange .
it was shown also that the shrinkage of the diffractive peak with increasing energy can be explained by the following pomeron trajectory , @xmath87 with the pomeron intercept @xmath88 and slope @xmath89 gev@xmath12 . at @xmath90 gev@xmath85 and large energy , however , the experimental data for the differential cross sections for elastic proton - proton and proton anti - proton scattering show a very different behavior .
the differential cross sections in this kinematical region do not show _ any _ energy dependence and are only functions of the momentum transfer @xmath76 .
( see fig .
[ fig : pp ] ) several explanations for the large @xmath71 data have been suggested .
the most popular one is the donnachie - landshoff odderon - exchange model based on a perturbative three - gluon exchange mechanism between nucleon quarks @xcite . in this model
, however , the average momentum transfer in each quark - quark subprocess in the experimental range of @xmath76 is rather small , @xmath91 gev , and therefore justification for its applicability is still unclear @xcite .
( see also ref .
the validity of the perturbative approach to qcd to explain nucleon - nucleon elastic scattering data is even less clear for the sotiropoulos - sterman model @xcite where a six - gluon effective exchange mechanism between nucleons was considered .
the new trajectory , which we relate to the effective @xmath25-exchange and to the manifestation of axial anomaly in the strong interaction , gives a more natural explanation to this experimental data . in this section , we calculate the contribution of the @xmath25-exchange to the differential cross section for proton - proton elastic scattering at large energy , @xmath92 and @xmath93 ; @xmath94 the lagrangian of the @xmath25-meson interaction with the nucleon for a point - like vertex is @xmath95 which allows the @xmath25-exchange to generate non - spin - flip amplitudes in nucleon - nucleon scattering .
the matrix element in eq .
( [ cross ] ) at @xmath96 reads @xmath97 ^ 4 s^2}{(t - m_{f_1}^2)^2 } , \label{matrix}\ ] ] which leads to the differential cross section of the elastic nucleon - nucleon scattering as @xmath98 ^ 4 } { 4\pi(t - m_{f_1}^2)^2}. \label{fcross}\ ] ] we note that it does _ not _ depend on the energy @xmath99 . in fig .
[ fig : pp ] , we present the contribution from the @xmath25-exchange to the differential cross section for proton - proton elastic scattering and compare with the experimental data at large energies , @xmath100 gev , @xmath101 gev , and @xmath102 gev , and large momentum transfer @xmath103 gev@xmath85 @xcite . in this calculation we have used @xmath104 , which was fixed by eq .
( [ gf1nn ] ) through the proton spin analysis .
the result shows that the @xmath25-exchange explains the experimental data on elastic proton - proton scattering at large momentum transfer very well . the pomeron contribution to the elastic @xmath26 scattering differential cross section @xcite , @xmath105 ^ 4}{4\pi } \left ( \frac{s}{s_0 } \right)^{2[\alpha_p^{}(t)-1 ] } \label{pcross}\ ] ] with @xmath106 gev@xmath107 and
@xmath108 gev@xmath85 is small at large @xmath71 compared with the @xmath25-exchange and can be neglected in this kinematical region .
of course at large energy the meson - exchange should be reggeized due to the contributions from the mesons of higher spin , @xmath109 , which belong to the same trajectory . as discussed above , however , there is a difference in the characteristic scales for the pomeron and the new trajectory . in the former ,
the scale is determined by the interactions between quarks at the distance related to the confinement radius , while in the latter it is related to the axial charge distribution ( [ rho ] ) .
therefore the two trajectories should have different slopes and @xmath110 the flatness of the new trajectory leads to small contributions of the heavier mesons to the cross sections .
for example , with the slope ( [ slope ] ) the next meson from the new trajectory with @xmath111 should have a mass @xmath112 gev as determined from the equation , @xmath113 in order to obtain a reasonable estimate we can safely neglect the contributions from the heavier mesons and take the contribution of the new trajectory as a fixed pole with @xmath114 .
the quantum numbers of this new trajectory are determined by the quantum numbers of the @xmath25 meson : the signature is @xmath115 with the parities @xmath4 .
thus the new trajectory should have the same contribution to @xmath26 and @xmath116 collisions .
this property is different from the charge odd odderon - exchange .
furthermore , the strength of the interaction of the new exchange with some hadronic state is determined by the value of the flavor singlet axial vector charge of the hadron .
this leads to the vanishing of the contributions of this trajectory to the total cross sections of the elastic reactions @xmath117 , @xmath118 , @xmath119 , etc , because the axial vector charge of pseudoscalar mesons is zero . one further consequence of the dominance of the new trajectory is that at large @xmath71 the elastic @xmath120 cross section , that is determined by the pomeron - exchange , should have very different @xmath76 and @xmath99 dependence in comparison with the @xmath26 case .
the experimental data @xcite support this conclusion .
one of the features of this new trajectory is that it is responsible for the _ spin dependence _ of the lepton - hadron , photon - hadron , and hadron - hadron cross sections at large energies .
thus the exchange of this trajectory should determine the behavior of the flavor singlet part of the spin - dependent structure function @xmath121 at @xmath122 : @xmath123 where @xmath124 .
similar behavior with @xmath125 for the neutron structure function was found by e154 collaboration @xcite .
the peculiarity of the neutron structure function @xmath126 is the smallness of the valence quark contribution .
therefore the behavior of this function is determined by the flavor singlet contribution in the experimentally available range of @xmath60 . for the proton target
the valence quark contribution is very large and the behavior in @xmath60 follows the @xmath70 trajectory contribution in this range of @xmath60 .
thus the neutron target is more suitable to investigate the new trajectory in polarized dis .
the recent hera data for vector - meson photoproduction @xcite are now the subject for discussion in the different approaches @xcite .
the interest to these data is related to the possible check of different nonperturbative and perturbative qcd models for particle photoproduction . in spite of the success of several models in the description of some properties of vector - meson photoproduction ,
a complete theory is not available so far .
indeed , the new zeus data for @xmath2 , @xmath3 , and @xmath127 photoproduction @xcite show that the perturbative approach may describe only @xmath127 photoproduction .
we also note that the nonperturbative approaches based on the dominant contribution of the _ universal _ pomeron trajectory fail to explain the cross sections at large momentum transfer @xmath71 . furthermore , the mechanisms which are responsible for the specific polarization properties of the electromagnetic production of vector mesons has not been clarified until now .
for example , it is very hard for such models to explain the large double spin asymmetry in @xmath2 meson electroproduction at relatively small value of @xmath128 observed recently by the hermes collaboration @xcite .
the specific features of the new trajectory can be extracted from the experimental data of vector - meson production . as we discussed above , because of the different @xmath76-dependence of the form factors in the pomeron - nucleon and @xmath25-nucleon vertices this new trajectory should dominate over the pomeron contribution at large @xmath71 region . to demonstrate the effect of the new trajectory we present here the numerical results for @xmath24 and @xmath3 photoproduction at large energy @xmath129 , where @xmath130 is the center of mass energy of the photon - nucleon system .
( definitions of the kinematical variables can be found in fig .
[ fig : vmp ] . )
we consider contributions from the two processes shown in fig .
[ fig : vmp ] .
the contributions from @xmath131 and @xmath132 exchanges which give corrections to the diffractive process at low energy @xcite are suppressed at large energy and are not considered in this work .
it is very well - known that the @xmath76-channel exchange by the pomeron trajectory gives the main contribution to the cross sections of vector - meson photoproduction at small @xmath71 .
the matrix element of this exchange within the donnachie - landshoff model reads @xcite @xmath133 \bar{u}_{m ' } ( p ' ) \gamma_\mu u_m(p ) \varepsilon^\mu_\gamma ( \lambda_\gamma ) \biggr\},\end{aligned}\ ] ] where the vector - meson and photon helicities are denoted by @xmath134 and @xmath135 while @xmath136 and @xmath137 are the spin projections of the initial and final nucleon , respectively , @xmath138 \right\},\ ] ] with @xmath139 , and @xmath140 is given in eq .
( [ f1 t ] ) .
the vector - meson mass is @xmath141 and @xmath142 is the pomeron trajectory given by eq .
( [ ptraj ] ) with @xmath143 and @xmath144 gev@xmath12 .
other parameters are @xmath145 and the vector - meson decay constants are @xmath146 and @xmath147 .
the vertex for the coupling of the axial vector current ( with momentum @xmath148 and polarization vector @xmath149 ) with two vector currents ( with momentum @xmath150 and polarization vectors @xmath151 ) , i.e. @xmath152 , can be described by two form factors @xcite , @xmath153 , \label{vertex}\end{aligned}\ ] ] where @xmath154 .
note that in general there are six form factors and two of them do not contribute to the physical processes . here
we follow the prescription of ref .
therefore , for vector - meson photoproduction by the exchange of axial vector current , the structure of the vertex becomes very simple in the local limit with constant @xmath155 : @xmath156 where @xmath157 .
this corresponds to the @xmath158 interaction lagrangian of ref .
@xcite obtained by using the hidden gauge approach .
the coupling constant @xmath159 can be determined from the experimental data on @xmath160 decay @xcite : @xmath161 mev and @xmath162 mev .
the vertex form of eq .
( [ vertex1 ] ) gives @xmath163 which leads to latexmath:[\[|g_{f_1 \rho^0 \gamma}| = 0.94 \mbox { gev}^{-2 } , \qquad therefore the matrix element of the @xmath25 meson exchange to elastic vector - meson photoproduction reads @xmath165 for the form factor of the @xmath166 vertex , @xmath167 , we take @xmath168 with @xmath169 gev and @xmath170 gev .
the results for the pomeron- and @xmath25-exchange contributions to the total cross sections of @xmath2 and @xmath3 meson photoproduction are presented in figs .
[ fig : rho ] and [ fig : phi ] .
one can find that the @xmath25-exchange contribution is nearly constant with the energy and its magnitude is at the level of a few percent , say @xmath171% , of that of the pomeron - exchange .
figures [ fig : rhot ] and [ fig : phit ] show the differential cross sections of @xmath2 and @xmath3 meson photoproduction at @xmath172 gev .
these show that the contribution of the @xmath25-exchange is very large and dominates over the pomeron contribution in the large @xmath71 region .
therefore , although the @xmath25 contribution is suppressed in the total cross section , its role can be found in the differential cross sections .
the generic structure of the new trajectory can also be found from the spin dependence of the related vertices , which is very different from that of the pomeron - exchange .
the main difference is that this exchange has unnatural parity , i.e. , @xmath173 .
one expectation is then that the contribution of this new trajectory may be separated by measuring spin observables , for example , the double spin asymmetries in hadron - hadron , lepton - hadron , and photon - hadron interactions , which vanish in the natural parity pomeron - exchange .
the wide experimental program for investigating polarization effects in proton - proton scattering has been suggested for rhic @xcite .
the main goal of this program is to extract information about polarized parton distribution functions of the nucleon and to check the important role of the axial anomaly in polarization physics .
we would like to emphasize that the polarized proton - proton _ elastic _ scattering could be very useful to extract the information about axial anomaly effects related with the contribution of the new trajectory .
as an example , we consider the effects of the @xmath25-exchange in the double longitudinal spin asymmetry in the @xmath26 elastic scattering : @xmath174 where @xmath175 denotes the differential cross section of the proton - proton scattering and arrows show the relative orientation of the proton spins
. this asymmetry can be written through the helicity amplitudes @xcite , @xmath176 as @xmath177 it is very well - known that at large energy @xmath178 one can neglect the contribution to the cross sections from the spin - flip amplitudes , @xmath179 , @xmath180 , and @xmath181 .
furthermore , the pomeron- and @xmath25-exchanges have different relations for the @xmath182 and @xmath183 amplitudes : @xmath184 therefore only the interference between the two exchanges can lead to non - vanishing @xmath185 @xmath186 where we have neglected the contribution from the imaginary part of the @xmath25-exchange to @xmath185 because of the very small deviation of the intercept of new trajectory from one . without this interference the asymmetry is suppressed as @xmath187 .
the final result for the asymmetry can then be written simply as @xmath188 \right\ } , \label{as4}\ ] ] where @xmath189 is the contribution from the pomeron- and @xmath25-exchange , respectively , to the differential cross section .
the result of the calculation for the two rhic energies , @xmath190 gev and @xmath191 gev is presented in fig .
[ fig : allpp ] .
this shows the large asymmetry at @xmath192 gev@xmath85 and the weak dependence on the energy , which is related to the high intercept of the new trajectory .
another very useful quantity which is also sensitive to the new trajectory is the difference of the total cross sections of polarized initial states : @xmath193 this asymmetry can be written in the terms of the helicity amplitudes , @xmath194 , \label{total1}\ ] ] using the optical theorem .
it should be mentioned that only unnatural parity exchange can contribute to @xmath195 .
since the pomeron - exchange has @xmath196 , it does not contribute to @xmath195 although it dominates in the unpolarized total cross section .
thus at large energy only the new anomalous trajectory which has unnatural parity can contribute . since the asymmetry @xmath195 is proportional to the imaginary part of the new trajectory , it vanishes if the intercept of the @xmath25 trajectory is @xmath197 exactly .
therefore the measurement of this asymmetry at large energy gives the opportunity to measure the deviation of @xmath198 from one : @xmath199
\right\}. \label{sec3}\ ] ] recently the e581/704 collaboration @xcite has reported a measurement on @xmath195 for proton - proton scattering at rather large momentum @xmath200 gev@xmath201 : @xmath202 from this result , with assumption that at this energy the contribution of other regge trajectories with unnatural parity is small , we can estimate the intercept of the new trajectory as @xmath203 which is in good agreement with eq .
( [ dis ] ) , which was extracted from the analysis on the behavior of the structure function @xmath126 in polarized deep - inelastic scattering at low @xmath60 . another suitable field to look for the effects of this trajectory is the polarized vector - meson photoproduction . given in fig .
[ fig : polbt](a ) are the results of the double spin asymmetries in @xmath2 and @xmath3 photoproduction for longitudinally polarized proton targets and polarized photon beams at @xmath204 gev .
we define the beam - target double asymmetry as @xmath205 where the arrows denote relative orientations of the proton spin and photon helicity . in fig .
[ fig : polbt](b ) we also give @xmath206 defined as @xmath207 where @xmath208 one can find that the values of @xmath209 and @xmath210 are very different from those of the pomeron - exchange model .
the @xmath25-exchange contribution gives rise to nonvanishing values for @xmath209 at small @xmath71 region , while the pomeron model predicts @xmath211 at this region .
it should be mentioned that recently the first measurements of the double spin asymmetries for vector - meson electroproduction was reported by the hermes collaboration @xcite .
they found large asymmetry for @xmath2 meson electroproduction with @xmath212 gev@xmath85 .
the asymmetries of @xmath3 and @xmath127 electroproduction could not be measured accurately and can not give any definite conclusion to these asymmetries .
nevertheless , the data on the double spin asymmetry of @xmath2 electroproduction could not be explained by the pomeron - exchange model and we propose the @xmath25-exchange as a possible candidate which gives nonvanishing double spin asymmetries at high energy .
we have shown that the @xmath5 meson plays a very special role in some hadronic processes in the large energy and large momentum transfer region due to its special relation to the axial anomaly through the matrix elements of the axial vector current .
this behavior is reminiscent of another effective particle with a very dominant behavior in hadron physics in this region , the pomeron , whose relation with the conformal anomaly seems to be the motivation behind this behavior .
therefore it seems natural to describe the properties of the lightest @xmath25 in terms of regge theory by associating to this resonance an anomalous trajectory , the odd signature companion of the even signature pomeron .
our analysis of proton - proton scattering and vector - meson photoproduction has confirmed our suspicions .
the contribution of the @xmath25-exchange to the cross sections of these processes does not depend on the energy , a clear signature of its anomalous regge behavior .
we have found that the @xmath25-meson exchange can explain the differential cross sections of proton - proton scattering at large energy and large momentum transfer .
the difference in the spin structure of the vertices of the @xmath25-trajectory from the pomeron - exchange leads to their different contributions to the polarized proton - proton scattering and polarized vector - meson photoproduction . as a result
, this trajectory gives non - zero values for the beam - target double spin asymmetries in both processes , and can be distinguished from the pomeron - exchange model .
these examples show the importance of the new trajectory in hadron reactions , and can be tested , for example , by the @xmath213 experiment at rhic @xcite , where a wide program to measure various spin - dependent elastic and total cross sections has been suggested .
( see recent discussions in ref . @xcite . ) as another test of the new trajectory , we suggest to investigate @xmath214 photoproduction , which decays mostly into @xmath215 . since the @xmath216 has quantum numbers @xmath217 , the empirical gribov - morrison rule @xcite prohibits the pomeron - exchange in this process , and because of the @xmath218 parity vector - meson - exchanges such as @xmath219 and @xmath2 can not contribute .
the possible pseudoscalar - meson exchange contribution , however , gives usually decreasing total cross section with the initial energy , while experiments @xcite show approximately constant total cross section at large energies .
therefore , we expect that the new trajectory related with the @xmath25 may give important contribution to @xmath216 photoproduction .
but the currently available data on this reaction @xcite are very limited and new experiments at current electron facilities are strongly called for . we also suggest to analyze the asymmetry @xmath220 of the two - meson decays of vector - mesons produced by linearly polarized photon beams @xcite . since this asymmetry is @xmath221 for natural parity exchange and @xmath222 for unnatural parity exchange , the @xmath25 contribution would be found from deviations of this quantity from @xmath221 at large energy . in summary , in the energy region of the analyzed experiments two trajectories ,
the pomeron and the @xmath25 , dominate the total cross sections and the asymmetries . and ,
therefore , this new anomalous regge trajectory , which is responsible for spin effects at large energy , should be considered as a natural odd signature companion to the pomeron .
we are grateful to a. e. dorokhov , s. b. gerasimov , e. a. kuraev , e. leader , and t. l. trueman for illuminating discussions .
y.o . is grateful to m. tytgat for useful informations .
d .- p.m . and
y.o . were supported in part by the kosef through the ctp of seoul national university .
v.v . was supported by dgicyt - pb97 - 1227 , erb fmrx - ct96 - 008 , and the theory division at cern , where part of this work was done .
g. altarelli , r. d. ball , s. forte , and g. ridolfi , acta phys . pol .
b * 29 * , 1145 ( 1998 ) ; b. lampe and e. reya , max planck inst .
report no .
mpi - pht-98 - 23 , hep - ph/9810270 ; e. leader , a. v. sidorov , and d. b. stamenov , phys
d * 58 * , 114028 ( 1998 ) , phys .
b * 445 * , 232 ( 1998 ) ; s. scopetta , v. vento , and m. traini , phys .
b * 421 * , 64 ( 1998 ) , phys .
b * 442 * , 28 ( 1998 ) ; p. faccioli , m. traini , and v. vento , nucl . phys . *
a656 * , 400 ( 1999 ) .
zeus collaboration , m. derrick _ et al .
_ , z. phys .
c * 63 * , 391 ( 1994 ) ; z. phys .
c * 69 * , 39 ( 1995 ) ; phys .
b * 377 * , 259 ( 1996 ) ; zeus collaboration , j. breitweg _ et al .
j. c * 1 * , 109 ( 1998 ) ; h1 collaboration , s. aid _ et al . _ ,
b463 * , 3 ( 1996 ) .
a. donnachie and p. v. landshoff , nucl
b311 * , 509 ( 1989 ) ; a. schfer , l. mankiewicz , and o. nachtmann , phys .
b * 272 * , 419 ( 1991 ) ; j. r. cudell , nucl . phys . *
b336 * , 1 ( 1990 ) ; m. g. ryskin , z. phys .
c * 57 * , 89 ( 1993 ) ; s. j. brodsky _
et al . _ ,
d * 50 * , 3134 ( 1994 ) ; t. arens , o. nachtmann , m. diehl , and p. v. landshoff , z. phys .
c * 74 * , 651 ( 1997 ) ; d. yu .
ivanov and r. kirschner , phys .
d * 58 * , 114026 ( 1998 ) ; i. royen and j .- r .
gudell , nucl . phys .
* b545 * , 505 ( 1999 ) ; l. l. frankfurt , m. f. mcdermott , and m. strikman , jhep * 9902 * , 002 ( 1999 ) ; s. v. goloskokov , eur .
j. c * 11 * , 309 ( 1999 ) ; s. i. manayenkov , desy report no . desy-99 - 016 , hep - ph/9903405 ( 1999 ) . a. i. titov , y. oh , and s. n. yang , phys .
* 79 * , 1634 ( 1997 ) ; a. i. titov and y. oh , phys .
b * 422 * , 33 ( 1998 ) ; a. i. titov , y. oh , s. n. yang , and t. morii , phys .
c * 58 * , 2429 ( 1998 ) ; y. oh , a. i. titov , s. n. yang , and t. morii , phys . lett .
b * 462 * , 23 ( 1999 ) . | we show that a new regge trajectory with @xmath0 and slope @xmath1 explains the features of hadron - hadron scattering and photoproduction of the @xmath2 and @xmath3 mesons at large energy and momentum transfer .
this trajectory with quantum numbers @xmath4 and odd signature can be considered as a natural partner of the pomeron which has even signature .
the odd signature of the new exchange leads to contributions to the spin - dependent cross sections , which do not vanish at large energy .
the links between the anomalous properties of this trajectory , the axial anomaly and the flavor singlet axial vector @xmath5 meson are discussed . |
let @xmath0 be a classical split real lie group of type @xmath1 , @xmath2 , or @xmath3 and let @xmath7 be its split real semi - simple lie algebra .
let @xmath8 be a maximal parabolic subgroup of @xmath0 of type @xmath9 given as the complementary set of a root @xmath10 of the dynkin diagram of @xmath7 .
the homogeneous space @xmath11 is called a minimal flag manifold of @xmath0 . in the context of type @xmath1 ( resp .
@xmath3 ) , @xmath11 may be realized as the orthogonal grassmannians , i.e. , submanifolds of isotropic subspaces in an odd ( resp .
even ) dimensional vector space equipped with a non - degenerate symmetric bilinear form . in the context of type @xmath2 , @xmath11
may be realized as the isotropic grassmannians , i.e. , submanifolds of isotropic subspaces in an even dimensional vector space equipped with a symplectic form . in this paper
we compute the cellular @xmath6-homology group of such manifolds developing the half - shifted diagrams that appears as a fundamental tool to describe the cellular structure given by the schubert varieties . in the general context of flag manifolds ,
the topology of the complex flag manifolds of semi - simple lie groups is well developed and also a very ramified area of research nowadays ( see brion @xcite ) .
yet the first results about the topology of real flag manifolds dates to 1970 s by burghelea hangan moscovici
verona @xcite . in the 1980 s , the work of duistermaat - kolk - varadarajan @xcite computes the mod 2 homology . only in the 1990 s a complete description of its integral homology was done by kocherlakota @xcite .
the morse theory appears as the main tool of these works .
the first explicit cellular approach was described by wiggerman @xcite to get the fundamental groups of the real flag manifolds .
those ideas where generalized in rabelo - san martin @xcite where a @xmath12-dimensional cellular version is obtained to improve the integral homology computations elucidating some ambiguities left in @xcite .
a more sophisticated point of view is the infinite dimensional representation theory of the real semi - simple lie group described by casian - stanton @xcite .
the euler characteristic was addressed in the paper by altomani
medori nacinovicht @xcite .
a complete description of the orientability was described by patro
san martin
seco @xcite .
notice that in the real context there are cells in all dimensions while in the complex one it appears only in even dimensions and hence torsion may occur in the real case whilst in the complex case the homology boundary maps are trivial elsewhere .
furthermore , the description of the cohomology ring over the integers of real flag manifolds is open in general .
some results have been done for the classical real grassmannians , i.e. , grassmannians of type a by casian - kodama @xcite . in this context , the homology groups were described by burghelea hangan moscovici
verona in the book @xcite .
now , if we consider other grassmannians ( isotropic and orthogonal ) , the only specific results are the computations of its euler characteristic ( @xcite ) and its orientability ( @xcite ) . in rabelo @xcite , we have a first step towards solving the problem , whose result was obtained only for the maximal case .
it is remarkable that the young diagrams has been largely used in the context of cohomology rings of complex grassmannians mainly because its proven product properties related to the comohology rings of such manifolds . instead ,
in the former paper of the second author @xcite , the strict young diagrams were the main tool to deal with homology of some real grassmannians . in this work ,
we extend those ideas about using adapted young diagrams to lead with the additive structure of the homology .
since the homology groups of real flag manifolds are completely described by what is called by ikeda - naruse as the beta - sequence of a weyl group element , one of the main contribution of this paper lies in the presentation of the beta - sequences of the weyl group elements that parametrizes all grassmannians of types b , c and d in terms of the half - shifted young diagrams . for type a ,
this was done by ikeda - naruse @xcite and also by graham - kreiman @xcite .
we also hope that these diagrams may provide a way to get some results about the cohomology in the same sense that was done by casian - kodama @xcite for type a. we now give a very brief introduction to these diagrams ( for details , see the sections [ secyoung ] and [ secroots ] ) and the results for homology .
let @xmath13 be the @xmath14-module freely generated by the schubert varieties @xmath15 , @xmath16 , the main discussion in this work is how to compute the integer coefficients @xmath17 of the boundary map @xmath18 . if @xmath19 , the coefficient has the formula @xmath20 , where @xmath21 is the integer defined by @xmath22 , @xmath23 and @xmath24 is the unique root such that @xmath25 . in general , a element @xmath16 , either of type b , c or d , can be written as a permutation of the form @xmath26 where @xmath27 , @xmath28 and @xmath29 .
using the definition in pragacz - ratajski @xcite , the set @xmath30 is bijective to set of double partitions @xmath31 and each double partition can be represented as a `` half '' shifted young diagram ( which for groups of type b and c , means that the young diagram of @xmath32 fits inside a @xmath33 square and @xmath34 fits inside a @xmath35 staircase shape ) .
the benefit to use the half shifted form for the young diagrams is demonstrated when we need to compute the set @xmath36 of all positive roots sent to negative by @xmath37 , which give us a easy way to get the coefficient @xmath38 only drawing its half shifted young diagrams .
given a schubert variety @xmath39 with partition @xmath40 such that @xmath19 , the diagram associated to @xmath41 is obtained by removing one box from the diagram of @xmath42 .
the partition @xmath41 is called 1 . a _ @xmath32-removing of @xmath42 _ if @xmath43 for some @xmath44 ; 2 . a _ @xmath34-removing of @xmath42 _ if @xmath45 for some @xmath46 .
our main theorem is stated below : given @xmath47 in @xmath30 such that @xmath48 , this can be either a @xmath32-removing or a @xmath34-removing of @xmath42 and consider @xmath49 the row where the box is removed .
the coefficient @xmath17 in the boundary map is obtained from @xmath50 , which depends on the group @xmath0 and which box is removed : @xmath51 the article is organized as follows .
section [ secpre ] presents the preliminaries about the flag manifolds and the bruhat decomposition .
section [ sechomflag ] summarizes the basic results about the cellular homology of real flag manifolds as obtained in @xcite .
section [ secyoung ] shows the description of the double young s diagram of all grassmannians and how it is possible to write the weyl element from the diagram .
section [ secroots ] presents the formulas to get the set @xmath36 of all positive roots sent to negative by @xmath37 .
section [ secmares ] states the main results describing the boundary maps of the cellular homology of grassmannians of type b , c and d. section [ secproofs ] explicitly presents the computations .
we thanks to david anderson for helpful suggestions and valuable discussions .
the main facts about semi - simple lie groups and their flag manifolds may be found in helgason @xcite , knapp @xcite , warner @xcite and san martin @xcite .
furthermore , @xcite is the reference for the notation and for some specific results is the article .
flag manifolds are defined as homogeneous spaces @xmath5 where @xmath0 is a non - compact semi - simple lie group and @xmath4 is a parabolic subgroup of @xmath0 .
let @xmath7 be a non - compact real semi - simple lie algebra .
the flag manifolds for the several groups @xmath0 with lie algebra @xmath7 are the same .
take a cartan decomposition @xmath52 and let @xmath53 be a maximal abelian sub - algebra contained in @xmath54 .
we denote by @xmath55 the set of roots of the pair @xmath56 and fix a simple system of roots @xmath57 .
denote by @xmath58 the set of positive and negative roots respectively and by @xmath59 the weyl chamber @xmath60 .
let @xmath61 be the direct sum of root spaces corresponding to the positive roots .
the iwasawa decomposition of @xmath7 is given by @xmath62 .
the notations @xmath63 and @xmath64 are used to indicate the connected subgroups whose lie algebras are @xmath65 and @xmath66 respectively .
a sub - algebra @xmath67 is said to be a cartan sub - algebra if @xmath68 is a cartan sub - algebra of @xmath69 .
if @xmath70 is a cartan sub - algebra of @xmath7 we say that @xmath7 is a split real form of @xmath69 . a minimal parabolic sub - algebra of @xmath7 is given by @xmath71 where @xmath72 is the centralizer of @xmath53 in @xmath65 .
let @xmath4 be the minimal parabolic subgroup with lie algebra @xmath73 .
note that @xmath4 is the normalizer of @xmath73 in @xmath0 .
we call @xmath74 the maximal flag manifold of @xmath0 and denote by @xmath75 the base point @xmath76 in @xmath5 . associated to a subset of simple roots @xmath77
there are several lie algebras and groups .
we write @xmath78 for the semi - simple lie algebra generated by @xmath79 , @xmath80 .
let @xmath81 be the connected group with lie algebra @xmath78 .
moreover , let @xmath82 be the sub - algebra generated by the roots spaces @xmath83 , @xmath84 and put @xmath85 .
the normalizer @xmath86 of @xmath87 in @xmath0 is a standard parabolic subgroup which contains @xmath4 .
the corresponding flag manifold @xmath88 is called a partial flag manifold of @xmath0 or flag manifold of type @xmath9 .
we denote by @xmath89 the base point @xmath90 in @xmath91 . in this paper
the classical split real lie groups considered are the orthogonal groups @xmath92 , @xmath93 , @xmath94 , @xmath95 , and the symplectic group @xmath96 , @xmath97 , where @xmath7 is the respective lie algebra given by @xmath98 , @xmath99 and @xmath100 . according to @xcite , @xmath98 , @xmath99 are real forms of type bdi and @xmath100 of type ci . given @xmath101
, we will work with the minimal flag manifolds of the form @xmath11 where @xmath102 is a maximal proper subset of the simple set of roots .
let @xmath103 be the set of simple roots of @xmath7 ordered as below : for each choice of group @xmath0 , the corresponding flag manifold has a geometric realization .
the orthogonal grassmannian of type @xmath1 is the set of @xmath104-dimensional isotropic subspaces in the vector space @xmath105 equipped with a nondegenerate symmetric bilinear form .
it will be denoted by @xmath106 .
the isotropic grassmannian of type @xmath2 is the set @xmath104-dimensional isotropic subspaces in the symplectic vector space @xmath107 .
this set will be denoted by @xmath108 .
the orthogonal grassmannian of type @xmath3 is the set of @xmath109-dimensional isotropic subspaces in the vector space @xmath110 equipped with a nondegenerate symmetric bilinear form .
it will be denoted by @xmath111 . a central role in our context
will be played by the weyl group @xmath112 associated to @xmath53 .
this is the finite group generated by the reflections over the root hyperplanes @xmath113 contained in @xmath53 , @xmath114 .
alternatively , it may be given as the quotient @xmath115 where @xmath116 and @xmath117 are respectively the normalizer and the centralizer of @xmath53 in @xmath63 ( the lie algebra of @xmath118 is @xmath72 ) .
viewing the elements of @xmath112 as product of simple reflections @xmath119 , @xmath120 , the length @xmath121 of @xmath122 is the number of simple reflections in any reduced decomposition of @xmath42 which is equal to the cardinality of @xmath123 , the set of positive roots sent to negative roots by @xmath37 .
if @xmath124 is a reduced decomposition of @xmath42 then we put @xmath125 it follows that @xmath126 .
this is called the @xmath24-sequence of @xmath42 .
there is a partial order in the weyl group called the bruhat - chevalley order .
we say that @xmath127 if given a reduced decomposition @xmath128 then @xmath129 for some @xmath130 . for a subset @xmath131 , the subgroup @xmath132 is defined to be the stabilizer of @xmath133 .
alternatively , @xmath134 may be seen as the subgroup of the weyl group generated by the reflections with respect to the roots @xmath84 .
we also define the subset @xmath135 of @xmath112 by @xmath136 since there exists a unique element @xmath137 of minimal length in each coset @xmath138 , @xmath135 is called the subset of minimal representatives of the cosets of @xmath139 in @xmath112 .
the bruhat decomposition presents the flag manifolds as a union of @xmath64-orbits , namely , @xmath140 where @xmath141 if @xmath142 .
each @xmath64-orbit through @xmath42 is diffeomorphic to a euclidean space .
such an orbit @xmath143 is called a bruhat cell .
its dimension is given by the formula @xmath144 where @xmath145 is the multiplicity of the root space @xmath146 and @xmath147 denotes the roots in @xmath148 generated by @xmath9 .
given any @xmath149 , in order to establish a relation between the dimension @xmath150 and the length @xmath121 , we must choose the minimal representative for @xmath42 in @xmath135 .
in this case , for @xmath151 , @xmath152 ( see @xcite , corollary 2.6 ) . in the case of a split real form ,
i.e. , @xmath153 and @xmath154 , it follows that @xmath155 a schubert variety is the closure of a bruhat cell , i.e. , @xmath156 for a maximal flag manifold we avoid the superscript @xmath157 and write @xmath158 the bruhat - chevalley order defines an order between the schubert varieties by @xmath159 a well known characterization of the schubert varieties is given by @xmath160
i.e. , schubert varieties endow the flag manifolds with a cellular decomposition .
we summarize the main results of @xcite . for our purpose ,
we are supposing that @xmath0 is a group of type b , c or d. consider firstly a maximal flag manifold @xmath161 . the cellular homology of a cw complex is defined from a cellular decomposition of the complex provided in our context by the schubert varieties . given a schubert variety @xmath162 , we fix once and for all reduced decompositions @xmath124 as a product of simple reflections , for each @xmath163 , and @xmath164 , @xmath120 .
let @xmath165 be the @xmath6-module freely generated by @xmath166 , @xmath163 .
the boundary maps @xmath167 are defined by @xmath168 for some coefficients @xmath169 . in case @xmath170
then @xmath171 . if @xmath172 then @xmath173 is computed as the degree of a map between cells of codimension one ( see @xcite , page 140 ) .
[ propappendix ] let @xmath174 .
the following statements are equivalent . 1
. @xmath175 and @xmath172 ; 2 .
if @xmath124 is a reduced decomposition of @xmath163 as a product of simple reflections , then @xmath176 is a reduced decomposition and @xmath177 .
the first main result is that the coefficient @xmath17 is the sum of the degree of two sphere homeomorphisms which has degree one . for any choice @xmath178 such that @xmath172 , the coefficient @xmath179 .
it is possible to get a more accurate expression for the coefficients @xmath17 in terms of roots . for @xmath124 and @xmath180 ,
reduced decompositions of @xmath42 and @xmath41 respectively ( see proposition [ propappendix ] ) , we define @xmath181 and @xmath182 [ teo : mainrs ] [ teoforcw ] let be @xmath183 be defined as in ( [ forsigmasoma ] ) . then @xmath184 for @xmath122 , let @xmath185 be the sum of roots in @xmath186 , i.e. , the sum of the @xmath24-sequences of @xmath42
. given @xmath42 and @xmath41 such that @xmath187 , let @xmath188 be the unique root ( not necessarily simple ) satisfying @xmath189 , that is , @xmath190 .
then it is possible to show that @xmath191 where @xmath192 is the sum ( [ forsigmasoma ] ) ( see @xcite , proposition 4.8 ) .
we get immediately the following formula for @xmath193 .
[ teoforcw1 ] @xmath194 where @xmath195 is the integer defined by @xmath196 and @xmath188 is the unique root such that @xmath189 .
this result was first obtained by kocherlakota ( see theorem a , @xcite ) by a morse homology approach .
the advantage here is working with reduced decompositions to compute the numbers @xmath197 . in the context of the partial flag manifolds @xmath198 , the schubert varieties are @xmath199 , @xmath200 .
[ teo : mainrs2 ] the cellular homology of @xmath198 is isomorphic to the homology of @xmath201 which is the boundary map of the free module @xmath202 generated by @xmath162 , @xmath203 , obtained by restricting @xmath204 and projecting onto @xmath202 . hence the coefficients @xmath205,[w^{\prime } ] ) $ ] for the boundary map @xmath206 of the cellular homology of the partial flag manifolds @xmath207 is @xmath208,[w^{\prime } ] ) = c(w , w^{\prime } ) .\ ] ] and the computation of @xmath205,[w'])$ ] reduces to a computation of @xmath17 on @xmath161 .
the goal of this section is to provide the main tool for dealing with the combinatorics of isotropic grassmannians .
the complexity of the weyl group that parametrizes the cellular decomposition of these manifolds has a very nice description in terms of what we call _ half shifted _ young diagrams .
the main source to construct such diagrams is the work of buch - kresch - tamvakis @xcite , where the model of @xmath209-strict partition and young diagrams is presented together with properties of grassmannians of all types .
another reference is given by ikeda - naruse @xcite and also by graham - kreiner @xcite in the context of maximal grassmannians where its found how to fill in each young diagram with the positive roots sent to negative by the respective weyl group element .
here we propose a model where it is generalized by showing how to fill in the diagrams defined by bkt for all types of grassmannians with the positive roots sent to negative by the respective weyl group element .
as far as we know , such a description has not been done yet elsewhere . in particular , this will be very useful to compute boundary maps for cellular homology .
consider the grassmannian @xmath210 which parametrizes @xmath104-dimensional isotropic subspaces of a real @xmath211-dimensional symplectic vector space .
let @xmath212 denote the set of integer partitions @xmath213 with @xmath214 so that the young diagrams of @xmath32 fits inside a @xmath215 rectangle .
also define @xmath216 the set of all strict partitions @xmath217 with @xmath218 . the weyl group @xmath219 for the root system @xmath220 are permutations in @xmath221 with a sign attached to each entry , that means it is isomorphic to the semidirect product @xmath222 ; we will write these elements as barred permutations of the form @xmath223 using the bar to denote a negative sign , and we take the natural order on them , as above . if @xmath224 is the parabolic subgroup generated by @xmath225 then the set @xmath226 of minimal length coset representatives of @xmath224 parametrizes the schubert varieties in @xmath210 . this indexing set @xmath227
can be identified by a set of barred permutations of the form @xmath228 where @xmath229 with @xmath230 , @xmath231 and @xmath232 .
there is a description of elements of @xmath30 by means of young diagrams , i.e. , each element @xmath233 corresponds to a pair of partitions @xmath31 where the `` top '' partition @xmath234 is defined slightly different from pragacz - ratajski @xcite @xmath235 for @xmath236 and @xmath237 ; the `` bottom '' partition is @xmath34 .
note that the length of @xmath42 is @xmath238 , where @xmath239 and @xmath240 .
the schubert varieties @xmath241 are then parametrized by the set @xmath242 of pairs @xmath243 with @xmath244 , @xmath229 and such that @xmath245 .
we may arrange the top and the bottom partition in the same diagram .
we propose here an alternative way of arrange the diagram of @xmath246 which corresponds just to a `` half '' shifted diagram in the sense presented by tamvakis @xcite .
we consider the top diagram @xmath32 left justified in the @xmath33 rectangle and , for each @xmath247 , the @xmath248-th row of @xmath34 is shifted to the right @xmath249 units . with that shift
, the bottom diagram may be seen inside a staircase partition and for reasons that will be clear soon we consider the staircase partition with @xmath12 rows .
the condition @xmath245 implies that the number of rows in the bottom diagram does not exceed the number of boxes in the last row of the top diagram .
let us denote by @xmath250 and by @xmath251 the set of square boxes with coordinates @xmath252 the top and the bottom ( shifted ) diagrams , respectively .
each diagram @xmath250 and @xmath253 is arranged in a plane with matrix - style coordinates .
the figure [ fig : schub_shift_554_8741 ] presents the half shifted diagram of the schubert variety @xmath254 in @xmath255 and how to read a @xmath256 box in each diagram . in @xmath255 . ] for each row in the top diagram , construct a diagonal line ( indicated as a dashed line in the figure [ fig : schub_related_554_8741 ] ) from the center of the last box at right of the respective row to the center of the first box in the first row of the bottom diagram ( it means that these boxes are @xmath209-related in this half shifted diagram as defined by tamvakis @xcite ) . consider the boxes of the staircase partition that are outside @xmath34 organized into columns .
the columns that contain a such @xmath209-related box are called _ related _ and the remaining columns are _ non - related_. the _ length _ of a column is the number of blank boxes of this column in the staircase @xmath35 shape .
we may recover the permutation associated to such diagram by taking the length of the related and non - related columns .
namely , the permutation element for @xmath257 is defined by @xmath258 in the equation , where @xmath259 are the length of the related columns , @xmath260 and @xmath261 are the length of the non - related columns .
for example , the partition @xmath262 corresponds to the element @xmath263 as illustrated in the figure [ fig : schub_related_554_8741 ] . in @xmath255 . ]
this description represents a slight modification with respect to that is given in @xcite in terms of the model of @xmath209-strict partitions .
the choice for representing schubert varieties by half shifted diagrams instead of a diagram associated to a @xmath209-strict partition will be clarified in the next section .
we now give a notion of row - reading expression for each @xmath264 , @xmath265 .
with respect to the top partition @xmath32 , let @xmath266 be defined , for @xmath267 and @xmath268 , by @xmath269 for a given partition @xmath244 associated to @xmath265 , the row - reading map is a bijection @xmath270 defined by assign the numbers increasingly to the boxes of @xmath250 from right to left starting from the bottom row to the top row .
then , we can form a word @xmath271 where @xmath272 , for all @xmath273 . with respect to the bottom partition @xmath34 ,
let @xmath274 , for @xmath275 , @xmath276 and @xmath277 , by @xmath278 as before , for a given partition @xmath229 associated to @xmath265 , the row - reading map is a bijection @xmath279 defined by assign the numbers increasingly to the boxes of @xmath250 from right to left starting from the bottom row to the top row . then
, we can form a word @xmath280 where @xmath281 , for all @xmath282 .
the concatenation of expressions and gives @xmath283 .
the maps @xmath284 , @xmath285 , @xmath286 and @xmath287 for the schubert variety given by the partition @xmath254 of @xmath255 are illustrated in figure [ fig : schub_rowreading_554_8741 ] .
then , @xmath288 and @xmath289 . in @xmath255 . ]
consider the odd orthogonal grassmannian @xmath290 which parametrizes @xmath104-dimensional isotropic subspaces of a real @xmath291-dimensional vector space equipped with a nondegenerate symmetric bilinear form .
the weyl group for the root system @xmath292 is the same as that for @xmath220 .
hence , they have the same diagrams
. the only distinction will appear in the level of roots .
this will be described in the next section .
consider the grassmannian @xmath294 which parametrizes @xmath109-dimensional isotropic subspaces of a real @xmath295-dimensional vector space equipped with a nondegenerate symmetric form .
we denote by @xmath296 the weyl group of type @xmath297 .
the elements of @xmath296 are signed permutations of @xmath298 with an even number of sign changes which will be denoted by barred numbers . if @xmath299 is the subgroup of @xmath296 generated by @xmath300 , then the set @xmath301 of minimal length coset representatives of @xmath299 parametrizes the schubert varieties in @xmath302 .
the set @xmath303 consists of barred permutations of the form @xmath304 where @xmath305 , @xmath306 , @xmath307 , @xmath308 and @xmath309 is equal to @xmath310 or @xmath311 so that the expression has an even number of bars .
the elements in @xmath303 correspond to a set @xmath312 of partition pairs , which involve a partition @xmath313 and a @xmath229 such that @xmath245 . more precisely ,
define the partition @xmath32 as @xmath314 for @xmath267 .
then , @xmath312 consists of * pairs @xmath315 such that @xmath316 ; these are elements of @xmath303 of the form @xmath317 which is referred as _ of type 0 _ ; * two types of pairs @xmath318 and @xmath319 $ ] such that @xmath320 ; these correspond to elements of @xmath303 of the form @xmath321 which is referred as _ of type 1 _ and @xmath322 } = ( \widehat{u}_{k } , \ldots , u_1 , \overline{\lambda_1 + 1 } , \ldots , \overline{\lambda_r+1 } , \overline{1 } , v_{n - k - r } , \ldots , v_1)\ ] ] which is referred as _ of type 2_. with such description we may represent these elements by means of young diagrams , i.e. , the pair of partitions @xmath323 $ ] ( where @xmath324 $ ] means a generic partition of any type ) provides the `` top '' partition @xmath32 and the `` bottom '' partition is @xmath34 .
the schubert varieties @xmath241 are then parametrized by the set @xmath312 of pairs @xmath325 $ ] with @xmath313 , @xmath229 and such that @xmath245 .
we also propose likewise in the type c case an alternative way of arrange the diagram of @xmath246 which corresponds just to a `` half '' shifted diagram in the sense presented by tamvakis ( see @xcite , page 19 ) .
we consider the top diagram @xmath32 left justified in the @xmath326 rectangle and , for each @xmath247 , the @xmath248-th row of @xmath34 is shifted to the right @xmath248 units . with that shift
, the bottom diagram may be seen inside a staircase partition with @xmath327 rows
. also denote by @xmath250 and by @xmath251 the set of square boxes with coordinates @xmath252 the top and the bottom ( shifted ) diagrams , respectively .
each diagram @xmath250 and @xmath253 is arranged in a plane with matrix - style coordinates .
for example , consider the schubert variety @xmath328 and @xmath329 in @xmath330 .
notice that @xmath331 which means that it is of type 0 .
the diagram and its respective half shifted diagram are illustrated in figure [ fig : schub_d_543_761 ] . in @xmath330 . ] for each row in the top diagram , construct a diagonal line ( indicated as a dashed line in the figure [ fig : schub_related_54x_763d ] ) from the center of the last box at right of the respective row to the center of the first box in the first row of the bottom diagram ( it means that these boxes are @xmath209-related in this half shifted diagram as defined by tamvakis @xcite ) .
consider the boxes of the staircase partition that are outside @xmath34 organized into columns .
the columns that contain a such @xmath209-related box are called _ related _ and the remaining columns are _ non - related_. the _ length _ of a column is the number of blank boxes of this column in the staircase @xmath35 shape .
we may recover the permutation associated to such diagram by taking the length of the related and non - related columns .
namely , the permutation element for @xmath332 $ ] is defined by @xmath258 in the equation , where @xmath259 are the length of the related columns , @xmath333 are the length of the non - related columns .
for example , the schubert variety @xmath334 of type 0 in @xmath330 corresponds to the permutation @xmath335 as illustrated in figure [ fig : schub_related_54x_763d ] ( left ) .
notice that one must fill in the diagonal boxes of the first @xmath336 rows that appear in the shifting procedure .
if the diagram is of type 0 , we have a unique diagram since @xmath316 .
if the diagram is of type 1 or 2 , since @xmath337 , we will have two possible diagrams : one of them with the diagonal box in the @xmath338-row not filled in and another with such box filled in . by definition
, they will correspond to the elements of type 1 and of type 2 , respectively .
consider , for example , the schubert varieties given by @xmath339 and @xmath340 $ ] of type 1 and 2 , respectively , inside @xmath330 as illustrated in figure [ fig : schub_related_54x_763d ] .
the permutations are , respectively , @xmath341 and @xmath342 .
( left ) , @xmath343 ( center ) and @xmath344 $ ] ( right ) in @xmath330 . ]
the diagrams presented here of type @xmath3 are also slight modifications of the model of the so called typed k - strict partitions presented by buck - kresch - tamvakis . for details , see @xcite , section 6 .
we notice that the diagrams presented here have the advantage of incorporating the @xmath345 factor that appears in the bkt - model . finally , we hope that the distinction between the type 1 and 2 weyl group elements were clarified in terms of its diagrams .
as it was made for groups of type c and b we have a row - reading expression for each @xmath346 , @xmath332 \in \mathcal{p}(k , n)$ ] .
we start with the maps @xmath347 that says how the boxes of a diagram are filled in with simple reflections .
* type 0 : * with respect to the top partition @xmath32 , let @xmath348 be defined , for @xmath267 and @xmath268 , by @xmath349 with respect to the bottom partition @xmath34 , let @xmath350 be defined , for @xmath275 , @xmath276 and @xmath277 , by @xmath351 * types 1 and 2 : * the case of types @xmath352 and @xmath353 also depends on the parity of the length of @xmath34 . with respect to the top partition @xmath32 , @xmath348 is defined , for @xmath267 and @xmath268 , by @xmath354 with respect to the bottom partition @xmath34 , @xmath355 is defined , for @xmath275 , @xmath276 and @xmath277 , by @xmath356 [ obs : t12 ] observe , in the equation , that for @xmath357 and @xmath358 , i.e. , for the first box at the last row of @xmath34 , the reflection is @xmath359 if @xmath42 has type 1 and @xmath360 if @xmath42 has type 2 .
then , we can easily find out the type of @xmath42 if we know which one appears first in the word : @xmath361 or @xmath362 , implying it is of type 1 or 2 , respectively .
now , we define the reading row map for all types @xmath363 and @xmath353 in the same way as made for groups of type c : define @xmath364 and @xmath365 by assign the numbers increasingly to the boxes of @xmath250 and @xmath251 , respectively , from right to left starting from the bottom row to the top row .
then , we can form the words @xmath366 @xmath272 and @xmath367 .
the concatenation of the equations gives @xmath368 .
the maps @xmath284 , @xmath285 , @xmath286 and @xmath287 for the schubert variety given by the partition of type 0 @xmath369 of @xmath330 are illustrated in the figure [ fig : schub_rowreading_543_763 ] . in @xmath330 . ] for the schubert variety given by the partition of type 1 @xmath370 and type @xmath353 @xmath371 $ ] of @xmath330 , this maps are described in the figure [ fig : schub_rowreading_544_763_t1_t2 ] . and @xmath371 $ ] in @xmath330 . ]
an important data needed to compute homology of groups is the description of the set @xmath372 of positive roots sent to negative roots by @xmath37 .
this section is devoted to show the relationship between @xmath372 and the respective half shifted young diagram of @xmath42 .
more specifically , since the number of roots in @xmath373 is equal to length @xmath121 , there is a rule that shows how each box in the diagram of @xmath42 is filled in with a root of @xmath373 .
our purpose is to describe the @xmath24-sequences of @xmath42 using the position @xmath256 of a box in the diagram of @xmath42 .
let @xmath241 be the schubert variety parametrized by @xmath243 with @xmath244 and @xmath229 .
for a box @xmath374 of @xmath34 define @xmath375 , where @xmath376 is the @xmath24-sequence for @xmath377 . for a box @xmath378 of @xmath32
define @xmath379 , where @xmath380 is the @xmath24-sequence for @xmath381 .
observe that @xmath382 is a permutation obtained by reordering @xmath42 according to the order of indexes given by @xmath383 .
this @xmath382 also can be seen as the element of the set @xmath384 of lagrangian grassmannians @xmath385 related to the shifted diagram @xmath34 .
[ propbetaseqbc ] the @xmath24-sequences can be described as follows : * the top part will be the same for both types c and b : @xmath386 * the bottom part is different for each case : 1 .
* type c : * @xmath387 ; 2 .
* type b : * @xmath388 ; + for @xmath374 by definition , @xmath389 is the @xmath24-sequence for @xmath377 and this formula is done in @xcite . to proof for the `` top '' part ,
it is enough to apply ( * ? ? ?
* lemma 3 ) to the definition of @xmath390 .
this proposition give us an easier way to compute the set @xmath36 of all @xmath24-sequences by placing such roots into each box of the young diagram .
[ ex : roots_554_8741 ] consider @xmath262 in @xmath255 . in this case , @xmath263 and @xmath391 .
first of all , we must label the rows and the columns of top diagram using @xmath42 , starting from the bottom row to the top and , afterwards , from the left column to the right .
then , the @xmath24-sequence associated to each box @xmath378 is the root @xmath392 , where @xmath393 and @xmath394 are the label of @xmath248-th row and @xmath395-th column , respectively .
for example , the box @xmath396 has labels 6 and @xmath397 for row and column , respectively , and it implies that @xmath398 as we can see in the figure [ fig : exemppioftypeb ] . for the bottom diagram , we must label the rows using the reflection of @xmath377 @xmath399 from the bottom row ( the @xmath12-th row at @xmath35 staircase ) to the top and using @xmath382 to label the columns from the left to the right .
the @xmath24-sequences are obtained likely to the top part . the figure [ fig : exemppioftypeb ] shows all the @xmath24-sequences of the partition @xmath246 .
-sequences associated to the partition @xmath254 . ]
let @xmath241 the schubert variety parametrized by pairs @xmath325 $ ] with @xmath313 , @xmath229 .
for a box @xmath374 of @xmath34 we define @xmath400 , where @xmath376 is the @xmath24-sequence for @xmath377 . for a box @xmath378 of @xmath32
we define @xmath401 , where @xmath380 is the @xmath24-sequence for @xmath381 .
we denote by @xmath402 the permutation associated to @xmath403}$ ] where we remove the hat from the first entry @xmath404 ( observe that @xmath402 might not lie in @xmath405 ) .
notice that @xmath382 is a permutation obtained by reordering @xmath42 according to the order of indexes given by @xmath406 .
this @xmath382 also can be seen as the element of the set @xmath384 of the grassmannian of @xmath407-dimensional isotropic subspaces @xmath408 related to the shifted diagram @xmath34 .
define @xmath409 by reordering @xmath402 according to the order of indexes @xmath406 .
[ propbetaseqd ] the @xmath24-sequences can be described as follows : * the top part will be given by : @xmath410 * the bottom part will be given by : @xmath411 considering each type separately , this proof if similar to proposition [ propbetaseqbc ] .
consider @xmath412 of type 1 and @xmath413 $ ] of type 2 in @xmath330 . in this case
, we have @xmath414 and @xmath415 ; @xmath416 and @xmath417 . repeating the same process as in the example [ ex : roots_554_8741 ] but now using the permutation @xmath402 to label the rows and columns instead of use the original @xmath42 ,
the figures [ fig : exemppioftypedi ] and [ fig : exemppioftypedii ] show the @xmath24-sequences for the partitions @xmath418 and @xmath419 , respectively .
-sequences of the partition @xmath370 of type @xmath352 . ]
-sequences of the partition @xmath371 $ ] of type @xmath353 . ]
let @xmath0 be either an orthogonal or a symplectic group and let @xmath8 be the parabolic subgroup , where @xmath9 is the complementary simple roots of @xmath420 of the respective dynkin diagram .
let @xmath165 be the @xmath14-module freely generated by the schubert varieties @xmath15 , @xmath421 .
the coefficients @xmath17 of the boundary map @xmath422 are defined by the equation .
the cellular homology of the isotropic grassmannian @xmath210 , odd and even orthogonal grassmannians @xmath290 and @xmath423 is obtained applying the theorem [ teo : mainrs ] which provides a formula to compute the coefficients @xmath17 .
this proof will be done in next section .
let @xmath15 be a schubert variety ( of any isotropic or orthogonal grassmannian ) and the associated double partition @xmath424 $ ] , where @xmath425 and @xmath426 which @xmath427 . given a schubert variety @xmath39 with partition @xmath40 such that @xmath187 , the partition @xmath41 is called 1 . a _ @xmath32-removing of @xmath42 _ if @xmath43 for some @xmath44 ; 2 .
a _ @xmath34-removing of @xmath42 _ if @xmath45 for some @xmath46 .
when we look at the diagram , removing a box from a row of the partition @xmath246 means that we are removing the last box of the respective row . in respect to the diagram , there are three different kinds of a @xmath34-removing of @xmath42 : 1 .
@xmath34-removing of @xmath42 _ from the diagonal _ ( only happens for type c and b ) ; 2 . @xmath34-removing of @xmath42 _ from a related column _ ; 3 . @xmath34-removing of @xmath42 _ from a non - related column _ ; observe that for any @xmath34-removing for groups of type c and b , by symmetry , we have @xmath428 unfilled boxes in the staircase pattern below the removed box .
being this a related column or not , @xmath429 or @xmath430 , is equal to @xmath428 .
the same idea holds for type d. this implies the following lemma : [ lemaremovnondiagbox ] suppose that @xmath41 is a @xmath34-removing of @xmath42 and @xmath431 for some @xmath46 .
recall the formulas , for type c and b , and , for type d , of @xmath42 in terms of the lengths @xmath432 s and @xmath433 s . *
if it is a @xmath34-removing from a related column , then @xmath434 * if it is @xmath34-removing from a non - related column , then @xmath435 [ teo : main ] given @xmath47 in @xmath30 such that @xmath48 , this can be either a @xmath32-removing or a @xmath34-removing of @xmath42 and consider @xmath49 the row where the box is removed .
the coefficient @xmath17 in the boundary map of the cellular homology of the respective @xmath11 is given by : @xmath436 where @xmath437 and @xmath50 depends on the group @xmath0 and which box is removed : @xmath438
therefore , we have four different possibilities to remove a box from @xmath246 as listed below . in these computations
we only need to know what happens to the top part since in @xcite we can find the proof for the bottom part .
remember that the permutation of @xmath42 can be written as @xmath441 and these values are obtained by counting the length of related and non - related columns of the bottom diagrams .
suppose that we are removing a box from the row @xmath442 for some @xmath443 .
looking through an imaginary diagonal line , the row @xmath442 is related to the @xmath444-th column in the bottom diagram @xmath253 .
clearly , the @xmath445-th column must be non - related .
after remove the box @xmath450 , the related @xmath444-th and non - related @xmath445-th columns of @xmath42 swicht their role in the permutation and they become , respectively , non - related and related columns of @xmath41
. then , the permutation of @xmath41 is obtained from @xmath42 just by swapping @xmath429 and @xmath430 @xmath451 denote by @xmath452 the @xmath24-sequences associated to @xmath41 . using proposition [ propbetaseqbc ]
, we can observe that the @xmath24-sequences in the top diagram are equal to @xmath42 and @xmath41 unless for those box at @xmath453-th row or @xmath442-th column . by definition ,
@xmath454 consider @xmath262 in @xmath255 and its @xmath24-sequences in figure [ fig : exemppioftypeb ] .
suppose that @xmath463 , where @xmath464 .
the @xmath24-sequences for @xmath41 can be described in the diagram as same as the ones for @xmath42 . comparing both diagrams
, we can see in figure [ fig : exempcase1c ] ( left ) that the @xmath24-sequences filled in the boxes are only different for some in the top part which have a dot .
subtracting the diagram of @xmath24-sequence of @xmath41 from the diagram of @xmath42 , we get an integer multiple of the root @xmath465 in the dotted boxes and zero in all other . in the figure [ fig : exempcase1c ] ( right )
, the multiplicity of such root is 1 for the removed box and for all in the same row or same column in the top part .
then , @xmath439 can be easily computed just using the diagram by adding all coefficients in it . for this example , @xmath466 . the above process can be applied in general for any @xmath32-removing of @xmath42 : put 1 s into all boxes of the top part in the same column and row of the removed box . adding all this numbers we get exactly the coefficient @xmath468 . although this whole work is devoted to establish the relation between the half - shifted young diagrams and grassmannians for groups of type b , c and d
, we can point out that the schubert varieties of the usual grassmannian ( of type a ) are paremetrized by the standard young diagram , a particular case of half - shifted when the bottom part @xmath470 .
then , the homology coefficients for the usual grassmannians can be computed considering only the @xmath32-removing case as made above .
suppose that we are removing a box from the row @xmath471 for some @xmath472 such that this is in the diagonal of the diagram @xmath253 of @xmath42 .
so , we must have @xmath473 and @xmath474 . since the new `` free '' of @xmath41 box is in the first non - related column , the permutations @xmath42 and @xmath41 are @xmath475 by proposition [ propbetaseqbc ]
, we can observe that the @xmath24-sequences in the top diagram are equal to @xmath42 and @xmath41 unless for those in the @xmath476-th column .
observe that , since @xmath427 then all the boxes in the column @xmath476 must belong to the diagram @xmath250 .
then , @xmath477 consider @xmath262 in @xmath255 ( type c ) and suppose that @xmath483 , where @xmath484 .
subtracting the diagram of @xmath24-sequence of @xmath41 from @xmath42 , we get an integer multiple of the root @xmath485 into the dotted boxes , which is given in the figure [ fig : exempcase2c ] ( center ) .
then , the sum of such multiplicities is @xmath486 .
if @xmath246 is in @xmath487 ( type b ) , the integer multiple of the root @xmath488 is given in the figure [ fig : exempcase2c ] ( right ) and @xmath489 .
the above process can be applied in general for any @xmath34-removing of a diagonal column of @xmath42 : for groups of type c , put 1 s into all boxes in the same column of the removed box , in both top and bottom parts . adding all this numbers we get exactly the coefficient @xmath490 . in the other hand ,
for groups of type b , put 1 only into the removed box and 2 s into all boxes above and in the same column of the removed box . adding all this numbers we get exactly the coefficient @xmath491 .
suppose that we are removing a box from the row @xmath471 for some @xmath472 such that this belongs to a related column of the diagram of @xmath42 .
using lemma [ lemaremovnondiagbox ] , there exist a @xmath492 such that @xmath493 . when the box @xmath494 is removed from the bottom diagram of @xmath42 , it will increase by 1 the length @xmath429 and the permutation @xmath41 is @xmath495 using proposition [ propbetaseqbc ] , we can observe that the @xmath24-sequences in the top diagram are equal to @xmath42 and @xmath41 unless for those box in @xmath496-th row or @xmath453-th column .
then , @xmath497 consider @xmath262 in @xmath255 and suppose that @xmath505 , where @xmath464 and @xmath506 . subtracting the diagram of @xmath24-sequence of @xmath41 from @xmath42 , we get an integer multiple of the root @xmath507 into the dotted boxes , which is given in the figure [ fig : exempcase3c ] ( center )
. then , the sum of such multiplicities is @xmath508 .
if @xmath246 is in @xmath487 ( type b ) , the integer multiple of the root @xmath507 is given in the figure [ fig : exempcase3c ] ( right ) and @xmath509 .
the above process can be applied in general for any @xmath34-removing of a related column of @xmath42 : for type c , fill the diagram in the same way ; for type b , just change the 2 in the diagram by 1 .
adding all this numbers we get exactly the coefficient @xmath511 for type c and @xmath512 for type b. using lemma [ lemaremovnondiagbox ] , there exist a @xmath513 such that @xmath514 . removing the box @xmath494 from the bottom diagram of @xmath42
, it will increase by 1 the length of @xmath430 and the permutation @xmath41 is @xmath515 where @xmath516 is in the position @xmath517 in the permutation .
now , if @xmath532 , we must analyze the diagram differently .
first of all , two any consecutive rows @xmath395 and @xmath533 are related , respectively , to the @xmath534-th and @xmath535-th columns in the bottom part .
there are exactly @xmath536 non - related columns between them .
so , we can relate each box @xmath537 , for @xmath538 , to a non - related column between @xmath534-th and @xmath535-th columns by drawing a diagonal line from the box @xmath537 to a first box in the @xmath539-th column .
the figure [ fig : proptransp ] shows this relation to non - related columns using dashed lines . in the other hand
, we can observe that each column @xmath453 of the top part , for @xmath540 , is related , by the above process , to some non - related column in the bottom part , and such column is exactly the @xmath541-th column of @xmath34 .
consider @xmath262 in @xmath255 and suppose that @xmath548 , where @xmath549 and @xmath550 .
subtracting the diagram of @xmath24-sequence of @xmath41 from @xmath42 , we get an integer multiple of the root @xmath551 into the dotted boxes , which is given in the figure [ fig : exempcase4c ] ( center ) .
then , the sum of such multiplicities is @xmath552 .
if @xmath246 is in @xmath487 ( type b ) , the integer multiple of the root @xmath551 is given in the figure [ fig : exempcase4c ] ( right ) and @xmath553 .
the above process can be applied in general for any @xmath34-removing of a non - related column of @xmath42 : for type c , fill the diagram in the same way ; for type b , just change the 2 in the diagram by 1 . adding all this numbers we get exactly the coefficient @xmath555 for type c and @xmath556 for type b. the boundary operator is obtained once we compute @xmath439 where @xmath440 is the sum of the elements in the @xmath24-sequence of @xmath42 .
the approach for the type @xmath3 is almost the same as for the types @xmath1 , @xmath2 . in some cases ,
the calculations are exactly the same . however , the existence of elements of type @xmath363 and @xmath353 asks for a careful analysis in some specific situations which turn out to be our focus in this section .
we use the results obtained by @xcite for the bottom part when needed and compute here the top part .
suppose that we are removing a box from the row @xmath442 for some @xmath443 .
we first observe that the same pattern over the diagrams verified in the types @xmath558 occur here , namely , this removing transforms a related column and a non - related column , respectively , into a non - related and related column .
explicitly , if @xmath429 , @xmath559 , and @xmath430 , @xmath560 , are the length of related column @xmath446 and non - related column @xmath447 , respectively , we have that @xmath561 observe also that @xmath448 and @xmath562 . in terms of the beta - sequences associated to the diagrams , the proposition [ propbetaseqd ] shows that the formula for the roots in the top diagram is the same as for the type @xmath558 given by the proposition [ propbetaseqbc ] . 1 .
if @xmath566 then we only have that @xmath563 , otherwise , if @xmath564 and @xmath566 then @xmath567 is minimal such that it is not possible to remove a box in the last row . since @xmath448 , we have that @xmath310 is unchanged which means that the type is preserved .
2 . if @xmath568 then we must have @xmath565 ( both @xmath563 and @xmath564 cases are included ) .
but since @xmath562 , @xmath565 implies that @xmath569 , i.e. , the term @xmath570 which characterizes the type of @xmath42 as @xmath352 or @xmath353 is preserved .
these cases are characterized by the fact that elements of type @xmath352 and @xmath353 are changed to a type @xmath573 element after removing the box . indeed ,
if @xmath574 @xmath575 is the partition for @xmath576 , with @xmath572 and @xmath260 , then @xmath577 will be the permutation given by the partition @xmath578 , i.e. , @xmath42 is an element of type @xmath579 and @xmath41 is an element of type @xmath573 .
the beta sequences of @xmath24 and @xmath452 of @xmath42 and @xmath41 , respectively , are represented in the corresponding diagram according to the proposition [ propbetaseqd ] .
we describe below only the boxes that are changed and contribute to the homology coefficient , namely , the @xmath209-row and the @xmath584-column of the top diagram . at these rows and columns
we have the following roots : @xmath585 implying that @xmath586 . for sake of illustration ,
let us consider the case when @xmath592 and @xmath336 is odd , i.e. , @xmath593 and @xmath594 .
there are @xmath476 roots @xmath595 given by the difference of roots at the @xmath476-th row of the top diagram .
there are also @xmath476 roots @xmath596 given by the difference of the roots at the @xmath476-column of the bottom diagram . adding them up we have @xmath476 roots @xmath597 which correspond to the same roots given by the difference of roots at the @xmath584-th column of the top diagram .
analogous considerations will hold when @xmath598 and @xmath336 is even .
first of all , consider @xmath412 in @xmath330 of type 1 and suppose that @xmath600 of type 0 , where @xmath549 . subtracting the diagram of @xmath24-sequence of @xmath601 from @xmath42
, we get : ( i ) for the top part , the sum of the roots is @xmath602 and ( ii ) for the bottom part , the sum of the roots is @xmath603 .
this implies that @xmath604 . in the other hand , considering @xmath413 $ ] of type 2 and the same @xmath605 , then ( i ) for the top part , the sum of the roots is @xmath606 and ( ii ) for the bottom part , the sum of the roots is @xmath607 , implying that @xmath608 . even though the above process does not give exactly the root @xmath611 as subtraction of @xmath24-sequences , the coefficient @xmath21 is given in the same way as the @xmath32-removing case for groups of type c and b : put 1 s into all boxes of the top part in the same column or row of the removed box . adding all this numbers we get exactly the coefficient @xmath468 .
suppose that we are removing a box from the row @xmath471 for some @xmath472 such that this belongs to a related column of @xmath42 .
by the lemma [ lemaremovnondiagbox ] , there is a @xmath612 such that @xmath613 .
lets considering two cases : when @xmath614 and when it is not .
after remove the unique box at the row @xmath616 , we have @xmath617 , which implies that @xmath41 does not have type 0 .
moreover , the simple permutation of the map @xmath285 at the removed box of @xmath42 depends on the parity of @xmath260 : it is @xmath361 if @xmath476 is even and @xmath362 if @xmath476 is odd .
then , the simple permutation at the first box of the row @xmath618 is @xmath362 if @xmath476 is even and @xmath361 if @xmath476 is odd , implying that @xmath41 has type 2 and 1 , respectively , by remark [ obs : t12 ] .
we can write the permutation of @xmath42 and @xmath41 as @xmath619 since this new `` free '' box is in the first related column and has length 2 . by proposition [ propbetaseqd ] , we can observe that the @xmath24-sequences in the top diagram are equal to @xmath42 and @xmath41 unless for those in the last row and in the @xmath476-th column .
therefore , @xmath620 then , @xmath621 .
now , the @xmath24-sequence for the bottom diagram change only for the @xmath476-th and @xmath584-th columns .
in fact , the permutations are @xmath622 and the @xmath24 sequences are : @xmath623 if @xmath566 then @xmath627 and @xmath41 still has type @xmath573 because remove a box from @xmath628 can not change the number of rows of @xmath629 . since @xmath630 , the permutations @xmath42 and @xmath41 are @xmath631 observe that the hat at the first element wo nt be a problem to compute the @xmath24-sequences using the proposition [ propbetaseqd ] because @xmath627 .
now , if @xmath632 or 2 then @xmath41 still has the same type of @xmath42 and the permutations are @xmath633 again , the hat at the first position wo nt affect the @xmath24-sequences because the proposition [ propbetaseqd ] uses , for type 1 and 2 , the permutations @xmath402 and @xmath634 , which are , by definition , the same permutations but without the hat at the first position .
consider @xmath334 in @xmath330 of type 0 and suppose that we are removing the element @xmath637 .
since the permutation of @xmath42 can be written , by the row - reading , as @xmath638 then @xmath639 and , by remark [ obs : t12 ] , @xmath41 has type 1 and it s partition is @xmath640 , where @xmath549 and @xmath641 .
then , we get : ( i ) for the top part , the sum of the roots is @xmath642 and ( ii ) for the bottom part , the sum of the roots is @xmath643 .
this implies that @xmath644 .
the above process gives exactly the root @xmath645 as a subtraction of @xmath24-sequences , the coefficient @xmath21 is given in the same way as the @xmath34-removing case of a related column for groups of type c and b : fill the diagram in the same way of above example .
adding all this numbers we get exactly the coefficient @xmath646 .
suppose that we are removing a box from the row @xmath471 for some @xmath472 such that this belongs to a non - related column of @xmath42 . by the lemma [ lemaremovnondiagbox ] ,
there is a @xmath647 such that @xmath648 .
lets considering two cases : when @xmath614 and when it is not .
this case means we are removing the unique box at the last row of @xmath34 and @xmath42 must have type 1 or 2 otherwise if @xmath42 has type 0 , the box lies in related column . clearly , @xmath473 and @xmath649 , as @xmath650 . by the row - reading map , if @xmath632 then @xmath651 .
in fact , @xmath632 implies that the removed box is labeled by @xmath359 and hence the diagonal of the last row of @xmath41 is labeled by @xmath360 , which means that @xmath651 .
for the same reason , if @xmath652 then @xmath653 . the permutation @xmath42 is @xmath654 and when we remove the box @xmath655 , the permutation of @xmath41 becomes @xmath656 by proposition [ propbetaseqd ] , we can observe that the @xmath24-sequences in the top diagram are equal to @xmath42 and @xmath41 unless for those in the @xmath476-th and @xmath584-th columns and both columns must have @xmath209 box each because @xmath657 . then @xmath658 then , @xmath659 .
now , the @xmath24-sequence for the bottom diagram change only for the @xmath476-th and @xmath584-th columns .
in fact , the permutations are @xmath660 and the @xmath24 sequences are : @xmath661 if @xmath566 then @xmath41 still has type @xmath573 because remove a box from @xmath628 can not change the number of rows of @xmath629 . since @xmath664 , the permutations @xmath42 and @xmath41 are @xmath665 now , if @xmath632 or 2 then @xmath666 and @xmath41 still has the same type of @xmath42 .
the permutations are @xmath667 where @xmath581 changes according to the type of @xmath42 .
observe that the no matter the type of @xmath42 , it will only change the values in @xmath668-th and @xmath669-th positions . to compute the @xmath24-sequences of the top and bottom parts
, we can use the same ideas as in case 4 of type c and the results in @xcite .
hence , @xmath670 first of all , consider @xmath412 in @xmath330 of type 1 and suppose that we are removing the element @xmath637 .
since the permutation of @xmath674 can be written , by the row - reading , as @xmath675 then @xmath676 and , by remark [ obs : t12 ] , @xmath601 has type 2 and it s partition is @xmath677 $ ] , where @xmath549 and @xmath678 .
we get : ( i ) for the top part , @xmath679 and ( ii ) the sum of the roots is @xmath680 .
this implies that @xmath681 .
in the other hand , considering @xmath413 $ ] of type 2 , the same argument shows that @xmath682 has type 1 and @xmath683 .
the above process gives exactly the root @xmath685 as subtraction of @xmath24-sequences , the coefficient @xmath21 is given in the same way as the @xmath34-removing case of a non - related column for groups of type c and b : fill the diagram in the same way of above example .
adding all this numbers we get exactly the coefficient @xmath512 . | let @xmath0 be a classical split real lie group of type @xmath1 , @xmath2 , or @xmath3 and @xmath4 a maximal parabolic subgroup of @xmath0 .
the homogeneous space @xmath5 is a minimal flag manifold of @xmath0 and may be realized , respectively , as an even orthogonal , isotropic and odd orthogonal real grassmannian . in this paper
we compute the cellular @xmath6-homology group of such manifolds by a lie theoretic approach which naturally furnish them with a cellular structure given by the schubert varieties .
the results are well understood in terms of what we call half - shifted young diagrams whose construction is based on the the works of buch - kresch - tamvakis , ikeda - naruse and graham - kreiman . |
in the recent years , new discoveries have been made at many high energy particle physics laboratories , including the @xmath22 , concerning physics within the boundaries of the standard model and beyond , as the consequence of pushing the maximum energy of the experiments to the new limits .
today , many of these laboratories use parton distribution functions ( @xmath13 ) to describe and analysis their extracted data from the deep inelastic @xmath23 collisions .
these scale - dependent functions are the solutions of the @xmath24-@xmath25-@xmath26-@xmath27-@xmath28 ( @xmath18 ) evolution equations , @xcite , @xmath29 , \label{eq1}\ ] ] where @xmath30 can be either the distribution function of the quarks , @xmath31 , or that of the gluons , @xmath32 , with @xmath33 being the fraction of the longitudinal momentum of the parent hadron ( the @xmath34 variable ) .
the terms on the right - hand side of the equation ( [ eq1 ] ) , correspond to the real emission and the virtual contributions , respectively .
the scale @xmath35 is an ultra - violet cutoff , related to the virtuality of the exchanged particle during the deep inelastic scattering ( @xmath36 ) .
@xmath37 are the splitting functions of the respective partons which account for the probability of emerging a parton @xmath38 from a parent parton @xmath39 through @xmath40 .
the @xmath18 evolution equation however , is based on the @xmath41 assumption , which systematically neglects the transverse momentum of the emitted partons along the evolution ladder .
it has been repeatedly hinted that undermining the contributions coming from the transverse momentum of the partons may severely harm the precision of the calculations , especially in the high energy processes in the small-@xmath33 region , see for example the references @xcite .
this signaled the necessity of introducing some transverse momentum dependent parton distribution functions ( @xmath42 ) , initially trough the @xmath43-@xmath44-@xmath45-@xmath46 ( @xmath47 ) equation @xcite , @xmath48 @xmath49 the @xmath50 implies a physical condition , enforcing the increase of the angle of the emission of the gluons in successive radiations along the evolution chain .
this condition which is usually referred to as the angular ordering constraint ( @xmath51 ) , is due to the coherent radiation of the gluons .
the @xmath52 form factor , @xmath53 , gives the probability of evolving from a scale @xmath54 to a scale @xmath55 , without any partons emission , and can be defined as : @xmath56 with @xmath57 . in the equation ( [ eq2 ] ) , @xmath58 is the double - scaled @xmath47 @xmath59 @xmath13 , which in addition to the @xmath33 and @xmath55 , depends on the transverse momentum of the incoming partons , @xmath8 .
it has been shown ( see the reference @xcite ) that in the proper boundaries , the @xmath47 equation will reduced to the conventional @xmath18 and @xmath60-@xmath61-@xmath62-@xmath26 ( @xmath20 ) equations , @xcite .
the procedure of solving the @xmath47 equation is mathematically involved and unrealistically time consuming , since it includes contemplating iterative integral equations with many terms . on the other hand , the main feature of the @xmath47 equation , i.e. the @xmath51 , can be exclusively used for the gluon evolution and therefore , this process is incapable of producing convincing quark contribution . to overcome these obstacles , martin et
al have introduced the @xmath8-factorization framework and developed the @xmath0-@xmath1-@xmath2 ( @xmath3 ) and the @xmath1-@xmath2-@xmath4 ( @xmath5 ) approaches @xcite , both of which are constructed around the @xmath21 @xmath18 evolution equations and modified with the different visualizations of the angular ordering constraint .
the frameworks of @xmath3 and @xmath5 in the @xmath21 and @xmath11 have been investigated intensely in the recent years , see the references @xcite .
although @xmath1 et al have developed the @xmath5 formalism as an improvement to the @xmath3 approach , by correcting the use of the @xmath51 , limiting its effect only on the diagonal splitting functions and extending the range of their calculations into the @xmath11 via introducing the @xmath10 scheme , it appears that the @xmath3 approach , as an effective model , is more successful in producing a realistic theory in order to describe the experiment .
we are therefore eager to expand our investigation regarding the merits and shortcomings of these frameworks into the calculation of the inclusive cross - sections of production of the electro - weak gauge bosons in high energy hadronic collisions .
the process of the production of the massive gauge vector bosons , @xmath63 and @xmath64 , have always been of extreme theoretical and experimental interest , since it can provide invaluable information about the nature of both the electro - weak and the strong interactions , setting a benchmark for testing the validity of the experiments and establishing a firm base for testing new theoretical frameworks , see the references @xcite .
it is not however straightforward to describe the transverse momentum distributions of the electro - weak bosons produced in hadron - hadron collisions , since the usual collinear factorization approach in the @xmath21 , neglects the transverse momentum dependency of the incoming partons and therefore predicts a vanishing transverse momentum for the product .
consequently , initial - state @xmath23 radiation is necessary to generate the @xmath8 distributions . on the other hand , in this approximation
, calculations for differential cross sections of the @xmath63 and @xmath64 production diverge logarithmically in the @xmath11 limit for the @xmath65 ( which is the main region of interest ) , due to the soft gluon emission .
so , one requires a re - summation to obtain a finite @xmath8 distribution . in the present work
we tend to calculate the @xmath8 distributions of the cross - section of production of the @xmath63 and @xmath64 using the @xmath11 level diagrams and the @xmath21 and @xmath11 @xmath7 of the @xmath3 and the @xmath5 frameworks .
the @xmath7 will be prepared in their proper @xmath8-factorization schemes using the @xmath13 of @xmath14 , @xmath15 , @xmath16 and @xmath17 , @xcite .
such calculations have been previously carried out using @xmath21 matrix elements of quark - antiquark annihilation cross section and doubly - unintegrated parton distribution functions ( @xmath66 ) in the framework of @xmath67-factorization , reference @xcite , and in a semi-@xmath11 approach , using a mixture of @xmath21 and @xmath11 matrix elements for the involved processes in addition to a variety of @xmath59 @xmath13 , see the reference @xcite .
to improve these approximations and at the same time , test the functionality of the @xmath3 and the @xmath5 @xmath7 , we have calculated the @xmath11 ladder diagrams for @xmath68 , @xmath69 and @xmath70 , utilizing a physical gauge for the gluons . in this way , at the price performing long and complicated calculations , we will demonstrate that with the use of the @xmath7 in the @xmath11 calculations , one can extract an excellent description of the experimental data of the @xmath71 [ 5,8,9 ] and @xmath72 [ 4 ] collaborations , as well as others works given here , regarding the transverse momentum distributions of the @xmath63 and @xmath64 boson .
in what follows , first , a brief introduction to the concept of @xmath8-factorization will be presented and the respective formalisms for the @xmath3 and the @xmath5 frameworks will be derived , in the section 2 . the section 3 contains a comprehensive description over the utilities and means for the calculation of the @xmath8-dependent cross - section of production of the @xmath63 and @xmath64 gauge vector bosons in a hadron - hadron ( or hadron - antihadron ) deep inelastic collision .
the necessary numerical analysis will be presented in the section 4 , after which a thoroughgoing conclusion will be followed in the section 5 .
a parton entering the sub - process at the top of the evolution ladder , has non - negligible transverse momentum .
however , it is customary to use the @xmath13 of the @xmath18 or the @xmath20 evolution equations to describe such partons , despite the fact that these density functions intrinsically carry no @xmath8-dependency . to include the contributions coming from the transverse momentum distributions of the partons
, one can either use the solutions of the @xmath47 evolution equation or unify the @xmath20 and the @xmath18 evolution equations to form a properly tuned @xmath8-dependent framework , @xcite . nevertheless , given the mathematical complexity of these schemes ,
it is not desirable to use them in the task of computing the @xmath36 cross - sections .
another way is to convolute the single - scaled solutions of the @xmath18 evolution equation and insert the required @xmath8-dependency via the process of @xmath8-factorization ( for a complete description see the reference @xcite ) .
thus , one may define the @xmath7 , @xmath73 , in the @xmath8-factorization scheme , through the following normalization relation , @xmath74 where @xmath75 are the solutions of the @xmath18 equation and stand for either @xmath76 or @xmath77 .
the procedure of deriving a direct expansion for @xmath73 , in terms of the @xmath13 is strait forward . yet , exposing the resulting prescriptions to the different visualizations of the @xmath51 will produce different @xmath7 , namely the @xmath3 , the @xmath9 and the @xmath10 frameworks . in what follows , we will describe these frameworks in detail .
starting form the @xmath18 equation in the leading order , the equation ( [ eq1 ] ) , and using the unregulated @xmath21 @xmath18 splitting kernels , @xmath37 , the reference @xcite , @xmath0 et al introduced an infrared cut - off , @xmath78 , as a visualization of the @xmath51 @xcite , @xmath79 limiting the upper boundary on @xmath80 integration by @xmath78 , excludes @xmath81 form the integral equation and automatically prevents facing the soft gluon singularities arising form the @xmath82 terms in the splitting functions .
additionally , they factorized the virtual contributions from the @xmath18 equations , by defining a virtual ( loop ) contributions as : @xmath83 with @xmath84 as an appropriated form of the @xmath52 form factor , the equation ( [ eq3 ] ) . afterwards , the double - scaled @xmath3 @xmath7 are defined as follows : @xmath85 .
\label{eq6}\ ] ] according to the above formulation , only at the last step of the evolution , the dependence on the second scale , @xmath86 , gets introduced into the @xmath7 .
the required @xmath13 is provided as input , using the libraries @xmath87 @xcite and @xmath88 @xcite , where the calculation of the single - scaled functions have been carried out using the @xmath36 data on the @xmath89 structure function of the proton .
@xmath90 are considered to be unity for @xmath91 .
this constraint and its interpretation in terms of the strong ordering condition gives the @xmath3 approach a smooth behavior over the small-@xmath33 region , which is generally governed by the @xmath20 evolution equation . in coordination with the theory of gluonic coherent radiation ,
it has been pointed out that the @xmath51 in the @xmath3 formalism should only act on the terms including the on - shell gluon emissions , i.e. the diagonal splitting functions @xmath92 and @xmath93 .
therefore , @xmath1 et al defined the @xmath21 @xmath5 @xmath7 as the correction to the @xmath3 framework @xcite , @xmath94 @xmath95 , \label{eq7}\ ] ] with @xmath96 for the quarks and @xmath97 @xmath98 , \label{eq9}\ ] ] with @xmath99 \right ) , \label{eq10}\ ] ] for the gluons . in the equations ( [ eq8 ] ) and ( [ eq10 ] ) , @xmath100 @xcite .
the @xmath7 of @xmath3 and @xmath5 to a good approximation , include the main kinematical effects involved in the @xmath36 processes .
one should note that the particular form of the @xmath51 in the @xmath3 formalism despite being of the @xmath21 , includes some contributions from the @xmath11 sector , whence in the case of @xmath5 framework , these contributions must be inserted separately .
the expansions of the @xmath21 @xmath5 formalism into the @xmath11 region can be achieved through the following definitions : @xmath101 @xmath102 with the @xmath11 splitting functions being defined as , @xmath103 and @xmath104 where @xmath105 stand for @xmath21 and @xmath11 respectively .
the reader can find a comprehensive description of the @xmath11 splitting functions in the references @xcite .
we must however emphasis that contrary to the @xmath3 and the @xmath21 @xmath5 frameworks , the @xmath51 is being introduced into the @xmath11 @xmath5 formalism via the @xmath106 constraint , in the `` extended '' splitting function .
now @xmath78 can be defined as : @xmath107 the @xmath11 corrections introduced into this framework are the collection of the @xmath11 @xmath13 , the @xmath11 splitting functions and the constraint @xmath108 .
nevertheless , it has been shown that using only the @xmath21 part of the extended splitting function , instead of the complete definition of equation ( [ eq12 ] ) , would result in reasonable accuracy in computation of the @xmath11 @xmath5 @xmath7 @xcite . additionally , the @xmath52 form factors in this framework are defined as : @xmath109 \right ) , \label{eq14}\ ] ] @xmath110 \right ) .
\label{eq15}\ ] ] each of these @xmath7 , the @xmath3 , @xmath21 and @xmath11 @xmath5 can be used to identify the probability of finding a parton of a given flavor , with the fraction @xmath33 of longitudinal momentum of the parent hadron , the transverse momentum @xmath8 in the scale @xmath86 at the semi - hard level of a particular @xmath36 process . in the following section
, we will describe the cross - section of the production of the @xmath63 and @xmath64 bosons with the help of our @xmath7 .
by definition , the total cross - section for a deep hadronic collision , @xmath111 , can be written in terms of its possible partonic constituents . utilizing the @xmath7 as density functions for the involved partons , one may write @xmath111 in the following form : @xmath112 @xmath113 where @xmath114 and @xmath115 are the incoming partons into the semi - hard process from the first and the second hadrons , respectively .
@xmath116 are the corresponding partonic cross - sections which can be defined separately as , @xmath117 @xmath118 and @xmath119 are the multi - particle phase space and the flux factor , respectively and can be defined according to the specifications of the partonic process , @xmath120 @xmath121 with the @xmath122 being the center of mass energy squared .
@xmath123 @xmath124 and @xmath125 are the 4-momenta of the incoming protons and since we are working in the infinite momentum frame , it is safe to neglect their masses .
@xmath118 can be characterized in terms of transverse momenta of the product particles , @xmath126 , their rapidities , @xmath127 , and the azimuthal angles of the emissions , @xmath128 , @xmath129 in the equation ( [ eq17 ] ) , @xmath130 are the matrix elements of the partonic diagrams which are involved in the production of the final results . to calculate these quantities
, one must first understand the exact kinematics that rule over the corresponding partonic processes .
the figure [ fig1 ] illustrates the ladder - type @xmath11 diagrams that one have to consider , counting the contributions coming from @xmath131 , @xmath132 , and @xmath133 as shown in the figure [ fig1 ] , panels ( a ) , ( b ) and ( c ) , respectively . the kinematics and calculations of this type of invariant amplitudes have been discussed extensively in the references @xcite .
we have followed the same approach , obtaining the @xmath134 terms only from the ladder - type diagrams , and not from the interference ( i.e. the non - ladder ) diagrams , using a physical gauge for the gluons , where only the two transverse polarizations propagate , @xmath135 @xmath136 is the gauge - fixing vector .
choosing such a gauge condition , ensures that the @xmath134 terms are being obtained from the ladder - type diagrams on both sides of the sub - processes . in the case of hadron - hadron collisions
, one might expect that neglecting the contributions coming from the non - ladder diagrams , i.e. the diagrams where the production of the electro - weak bosons is a by - product of the hadronic collision ( see the reference @xcite ) , would have a numerical effect on the results .
nevertheless , employing the gauge choice ( [ eq21 ] ) , one finds out that the contribution from the `` unfactorizable '' non - ladder diagrams vanishes . in the proton - antiproton center of mass frame , we can write the following kinematics @xmath137 @xmath138 where the @xmath139 are the 4-momenta of the partons that enter the semi - hard process . afterwards , it is possible to write the law of the transverse momentum conservation for the partonic process : @xmath140 with @xmath141 being the transverse momentum of the produced vector boson . additionally ,
defining the transverse mass of the produced virtual partons , @xmath142 , we can write , @xmath143 @xmath144 now , using the above equations , one can derive the following equation for the total cross - section of the production of the @xmath63 and @xmath64 bosons in the framework of @xmath8-factorization , @xmath145 @xmath146 @xmath147 note that the integration boundaries for @xmath148 are @xmath149 .
one may introduce an upper limit for these , say @xmath150 , several times larger than the scale @xmath86 , without any noticeable consequences .
yet , for @xmath151 with @xmath152 , i.e. for the non - perturbative region , it is impervious to decide how to validate our @xmath7 . a natural choice would be to fulfill the requirement that @xmath153 and therefore , we can safely choose the following approximation for the non - perturbative region : @xmath154 in the next section , we will introduce some of the numerical methods that have been used for the calculation of the @xmath155 , the equation ( [ eq25 ] ) , using the @xmath7 of @xmath3 and @xmath5 .
it it expected that through considering @xmath11 processes for this computation , the results will have a better agreement with the existing experimental data , in comparison with the previous calculations .
the main challenge one must face , in the computions of the total cross - section of a hadron - hadron collision in the @xmath11 , is the extremely complex calculations required for extracting the invariant amplitudes in a set of @xmath156 @xmath11 feynman diagrams .
each of our processes , @xmath68 , @xmath157 , and @xmath70 , include a number of different configurations , see the figure [ fig2 ] .
this is when we filter out the non - ladder diagrams , with our choice of the gauge condition on the gluon polarization , the equation ( [ eq21 ] ) .
writing the analytic expressions of the @xmath158 for theses diagrams is rather straight forward , see the appendix a. however , since the incoming and the out - going quarks are off - shell , and we do not neglect their transverse momenta , their on - shell spin density matrices has to be replaced with a more complicated expression . to do this , one can extend the original expressions , according to an approximation proposed in the references @xcite , through converting the off - shell quark lines to the internal lines via replacing the spinorial elements of the incoming and the out - going partons .
following this idea , we replace the incoming proton with a quark with the momentum @xmath159 and the mass @xmath160 which radiates a photon or a gluon and turns into an off - shell quark with the momentum @xmath161 .
therefore , the corresponding matrix element for such quarks can be written as , @xmath162 where @xmath163 represents the rest of the original matrix element .
now , the expression presented between @xmath163 and @xmath164 is considered to be the off - shell quark spin density matrix . using the on - shell identity @xmath165 and after performing some dirac algebra at the @xmath166 limit , one simply arrives to the following expression : @xmath167 \ ; \gamma^{\mu } \right).\ ] ] afterwards , imposing the sudakov decomposition @xmath168 with @xmath169 , one derives : @xmath170 thus , with the above replacement , the negative light - cone momentum fractions of the incoming partons have been neglected .
@xmath171 in this equation represents the properly normalized off - shell spin density matrix .
additionally , the coupling vertices of the off - shell gluons to quarks must be modified with the eikonal vertex ( i.e the @xmath20 prescription , see the reference @xcite ) .
therefore , in the case of initial off - shell gluons , we impose the so - called non - sense polarization condition , i.e. @xmath172 which results into the following normalization identity @xmath173 we can calculate the evolution of the traces of the matrix elements with the help of the algebraic manipulation system @xmath174 , @xcite .
also , the method of orthogonal amplitudes , see the reference @xcite , can be used to further simplify the results .
the numerical computation of the equation ( [ eq25 ] ) have been carried out using the @xmath175 algorithm in the monte - carlo integration .
to do this , we have selected the hard - scale of the @xmath7 to be equal to the transverse mass of the produced gauge vector boson : @xmath176 mathematically speaking , the upper bound on the transverse momentum integrations of the master equation ( 25 ) should be the infinity .
however , since the @xmath7 of @xmath3 and @xmath5 tend to quickly vanish in the @xmath177 domain , one can safely introduce an ultraviolet cut - off for these integrations . by convention , this cut - off
is considered to be at @xmath178 .
nevertheless , given that @xmath86 depends on the transverse momentum of the produced boson ( @xmath179 ) and its mass , it would be sufficient to set @xmath180 , with @xmath181 one can easily confirm that further domain have no contribution into our results . also it is satisfactory to bound the rapidity integrations to @xmath182 $ ] , since @xmath183 and according to the equation ( [ eq24 ] ) , further domain has no contribution into our results .
the choice of above hard scale is reasonable for the production of w and z bosons , as has been discussed in the reference @xcite . as a final note ,
we should make it clear that in the reference @xcite , the calculation of the transverse momentum distribution for the production of the @xmath184 and @xmath185 bosons has been carried out , using the aggregated contributions of the following sub - processes : * the @xmath11 @xmath186 partonic process , using the unintegrated gluon distributions of the @xmath47 and the @xmath21 @xmath5 formalisms , accounting for the production of the bosons accompanied by ( at least ) two distinct jets .
* the @xmath21 @xmath187 partonic process , with the density function of the incoming quarks and gluons being defined in the collinear ( @xmath188 or @xmath189 ) and the @xmath8-factorization ( the @xmath47 and the @xmath21 @xmath5 ) formalisms , respectively .
this corresponds to the @xmath190 cross - section . * the @xmath21 @xmath191 partonic process , from the collinear approximation , assuming that the incoming particles are valance quarks ( or valance anti - quarks ) .
the above paratonic processes ( a , b and c ) obviously neglect some of the @xmath11 contributions ( in the b and c cases ) , namely the shares of the non - valance quarks along the chain of evolution . additionally , assuming the non - zero transverse momentum for the valance quarks in the infinite momentum frame is to some extent unacceptable , since , in the absence of any extra structure , the intrinsic transverse momenta of the valance quarks should not be enough for producing the @xmath192 bosons with relatively large @xmath193 . in the present work ,
we have upgraded the partonic processes of the b and c cases with their nlo counterparts , i.e. @xmath194 and @xmath195 sub - processes .
so , we are able to use the updf of the @xmath8-factorization for the incoming quarks and gluons to insert the transverse momentum dependency of the produced bosons , and at the same time avoid over - counting . furthermore , the problem of separating the @xmath192+single - jet and the @xmath192+double - jet cross - sections will reduce to inserting the correct physical constraints on the dynamics of these processes , e.g. via inserting some transverse momentum cuts for the produced jets , using the anti-@xmath8 algorithm , see the reference @xcite .
nevertheless , since we are interested to calculate the inclusive cross - section for the production of the @xmath192 bosons , inserting such constraints is unnecessary .
using the theory and the notions of the previous sections , one can calculate the production rate of the @xmath63 and @xmath64 gauge vector bosons for the center - of - mass energy of @xmath196 @xmath197 .
the @xmath13 of martin et al @xcite , @xmath87 and @xmath88 , are used as the input functions to feed the equations ( [ eq6 ] ) , ( [ eq7 ] ) , ( [ eq9 ] ) and ( [ eq11 ] ) .
the results are the double - scale @xmath7 in the @xmath3 , the @xmath198 and the @xmath199 schemes .
these @xmath7 are in turn substituted into the equation ( [ eq25 ] ) to construct the @xmath192 cross - sections in their respective frameworks .
since we intend to compare our calculations to the @xmath200 and @xmath201 decays , we should multiply our theoretical out - put by the relevant branching fractions , i.e. @xmath202 and @xmath203 @xcite .
thus , the figures [ fig3 ] and [ fig4 ] present the reader , with a comparison between the different contributions into the differential cross - sections of the @xmath63 and @xmath64 , versus their transverse momentum ( @xmath8 ) in the @xmath3 scheme .
the main contributions into the production of the @xmath63 are those involving @xmath204 and @xmath205 vertices .
other production vertices have been calculated and proven to be negligible compared to these main contributions ( nevertheless , for the sake of completeness , we have included every single share , no matter how small they are in the total contributions , see the figures [ fig5 ] and [ fig6 ] , where the individual contributions of each of the production vertices in the partonic sub - processes for the production of @xmath63 and @xmath64 have been depicted clearly , in the framework of @xmath3 for @xmath206 ) . in the case of @xmath64 production ,
the main vertices are @xmath207 , @xmath208 , @xmath209 and @xmath210 . in both cases
, one can recognize the different behavior of various partonic sub - processes .
as expected , the contributions of the @xmath211 in all of the diagrams are similar , and even ( roughly ) of the same size , since they only depend on the behavior of the gluon density . on the other hand ,
the contribution coming from the @xmath212 differs from one production vertex to another , mimicking the differences between the quark densities of different flavors and going from the high contributions of the up and down quarks to small contributions of the charm and strange and even negligible contributions of the top and bottom quarks .
additionally , one notices the smallness of the @xmath213 contributions .
this is also anticipated , since the incoming gluon could ( with a relatively large probability ) decay into a quark - antiquark pair that does not have the right flavor to form a production vertex with considerable contribution .
the figures [ fig7 ] and [ fig8 ] illustrate a complete comparison between the results of the calculation of the production of the electro - weak gauge vector bosons in the frameworks of @xmath3 , @xmath198 and @xmath199 , with each other and with the experimental data of the @xmath71 and @xmath72 collaborations , references @xcite .
the results in the @xmath3 framework has an excellent agreement with the experimental data , both in the @xmath63 and @xmath64 productions .
the @xmath198 scheme behaves similarly compared to the @xmath3 framework , yet has a noticeably shorter peak , specially in the case of @xmath64 .
this is due the different visualization of the @xmath51 between these two frameworks , see the section 3 .
meanwhile , the results in the @xmath214 scheme are unexpectedly unable to describe the experiment data .
this is related to the conditions in which the @xmath51 has been imposed in this framework .
the @xmath215 constraint gives the parton distributions of the @xmath199 , a sharp descend to zero at @xmath216 and returns a vanishing contribution for the better part of the transverse momentum integration in the equation ( [ eq25 ] ) .
consequently , the overall value of the differential cross - sections of the @xmath63 and @xmath64 production in this framework reduces dramatically , as it is apparent in the figures [ fig7 ] and [ fig8 ] . in overall and as it has been stated elsewhere ( see for example the references @xcite ) the results in the @xmath3 scheme seemingly have a better agreement with the experiment .
this is to some extend ironic , since the @xmath21 and the @xmath199 formalisms are developed as extensions and improvements to the @xmath3 approach and are more compatible with the @xmath18 evolution equation .
such comparisons can also be made for the larger values of @xmath8 , see the figures [ fig9 ] and [ fig10 ] , where the production rates of the electro - weak gauge bosons are plotted against their transverse momentum for @xmath217 .
the diagrams include the calculations of @xmath218 and @xmath219 and the comparisons are made with the help of the data from the @xmath71 collaboration , references @xcite .
of course , since the data points have small values and large errors , and because of the closeness of the results in different frameworks , one can not stress over the superiority of any of the approaches . yet , our previous conclusion about the validity of the @xmath3 @xmath7 and the short - comings of the @xmath214 @xmath7 holds .
another interesting observation is that in the large @xmath8 , where because of the smallness of the results the higher order corrections become important , the calculations in the @xmath3 approach start to separate from the @xmath198 and behave similar to the @xmath199 .
the reason is that the inclusion of the @xmath220 splitting functions into the domain of the @xmath51 introduces some corrections from the @xmath11 region .
additionally , one notices that the contribution coming from the @xmath221 in the @xmath11 evaluations considerably deviates from the similar behavior of its respective counterparts .
this of course roots in the evolution of the @xmath11 quark densities in this framework , see the reference @xcite recently , martin et al have updated their @xmath13 libraries , the reference @xcite .
the figures [ fig11 ] and [ fig12 ] demonstrate the differences between the cross - section of the production of the @xmath192 vector bosons in the @xmath3 framework , using the ( older ) @xmath87 and the ( newer ) @xmath88 @xmath13 .
one notices that , using either of these @xmath13 as input for our @xmath7 produces a negligible difference .
the figures [ fig13 ] and [ fig14 ] present an interesting comparison between the experimental data and the results of the different approximations in the calculation of the production of the electro - weak gauge vector bosons .
in addition to our calculations in the @xmath3 and the @xmath5 @xmath7 in the @xmath21 and the @xmath11 approximations , the results coming from the @xmath47 @xmath59 @xmath13 ( reference @xcite ) , the @xmath222 @xmath6 parton distributions ( @xmath66 , see the reference @xcite ) and from the @xmath223 frameworks are included in these diagrams .
the @xmath47 results are calculated as the sum of @xmath224 , @xmath225 and @xmath226 sub - processes .
the @xmath66 results are in the @xmath227-factorization framework , utilizing a @xmath226 `` effective '' production vertex . furthermore , to calculate the differential cross - section of the @xmath192 production in the @xmath223 approximation , one have to ignore the transverse momentum integrations in the equation ( [ eq25 ] ) and replace the @xmath7 with the unpolarized parton distributions of @xmath87 , @xmath88 or @xmath228 @xcite : @xmath229 @xmath230 the reader should notice that the results of our computations in the @xmath11 regime , as expected , have a better behavior towards describing the experimental data , both in the @xmath63 and @xmath64 cases , since they descend with a shallow steep , compared to the results calculated in other schemes .
this is in part , because the @xmath11 evaluations are inherently more accurate . yet
, most of the credit goes to the precision of the utilized @xmath7 .
again , the @xmath3 framework in the @xmath11 calculations offers the best description of the experiment . additionally , it is possible to compare our presumed frameworks through the calculation of the total cross - section of the @xmath63 and @xmath64 production with respect to the center - of - mass energy of the hadronic collision , i.e. the figures [ fig15 ] and [ fig16 ] .
following our previous pattern , the results of both the @xmath3 and the @xmath198 frameworks show a good level of compatibility with the experimental data . on the other hand , since the @xmath199 framework has failed to describe the data , we have excluded its contributions here , to save some computation time .
finally , it has been brought to our attention that the @xmath231 and @xmath232 collaborations have recently published some data regarding the production of the @xmath64 gauge vector boson in the @xmath22 for @xmath233 , the references @xcite . in the above calculations ,
the rapidity of the produced boson has been separated in equally spaced rapidity sectors within @xmath234 domain . in figure
[ fig17 ] , we have addressed the above observations , using our @xmath11 framework and utilizing the @xmath7 of @xmath3 , since we have already established the superiority of this scheme in describing the experiment .
the individual contributions from the partonic sub - processes are presented and the total values of ( single and double ) differential cross - sections are subjected to comparison with the data of the @xmath231 and @xmath232 collaborations .
one easily notes that our calculations is in general agreement with the experimental data and with similar calculations in a @xmath235 @xmath23 framework from the reference @xcite .
unfortunately , performing these calculations are extremely time - consuming and the existing data points are not plentiful or accurate enough to let us make a decisive statement about the superiority regarding any of our presumed frameworks .
nevertheless , considering these comparisons , it is apparent that the @xmath3 @xmath7 in the framework of @xmath8-factorization , despite their miss - alignments with the theory of the @xmath18 evolution equation and the physics of the successive gluon radiations , as an effective theory , proposes the best option to describe the deep inelastic @xmath23 events .
however , until further phenomenological analysis , such claim remains as an educated speculation . in summary , within the present work
, we have calculated the rate of productions belonging to the electro - weak gauge vector bosons in the framework of @xmath8-factorization , utilizing the @xmath7 of @xmath3 , @xmath236 and @xmath199 , by the means of @xmath11 @xmath23 processes .
the results have been demonstrated and compared to each other and to the experimental data points from the @xmath71 and the @xmath72 collaborations , as well as the calculations in other frameworks . through our analysis we have suggested that despite the theoretical advantages of the @xmath5 formalism , the @xmath3 approach has a better behavior toward describing the experiment .
@xmath237 would like to acknowledge the research council of university of tehran and the institute for research and planning in higher education for the grants provided for him .
@xmath238 sincerely thanks a. lipatov and n. darvishi for their valuable discussions and comments .
@xmath238 extends his gratitude towards his kind hosts at the institute of nuclear physics , polish academy of science for their hospitality during his visit .
he also acknowledges the ministry of science , research and technology of iran that funded his visit .
given that we are interested in the calculation of the matrix element squared for each process , one immediately concludes that the @xmath239 . therefore it is sufficient to calculate the invariant amplitudes for the feynman diagrams of the figure [ fig2 ] , the panels ( b ) and ( c ) , which can be written as follows : @xmath240 with @xmath241 @xmath242 @xmath243 @xmath244 @xmath245 @xmath246 @xmath247 @xmath248 @xmath249 @xmath250 @xmath251 @xmath252 and @xmath253 @xmath254 @xmath255 @xmath256 @xmath257 @xmath258 @xmath259 @xmath260 @xmath261 @xmath262 @xmath263 @xmath264 @xmath265 @xmath266 @xmath267 @xmath268 where @xmath269 is the running coupling constant for @xmath23 and @xmath270 represents the vertex of the electro - weak gauge vector bosons with quarks : @xmath271 @xmath272.\ ] ] @xmath273 is the weinberg angle , @xmath274 is the corresponding @xmath275 matrix element and @xmath276 is the weak isospin component of the quark @xmath54 .
additionally , the standard @xmath23 three - gluon coupling can be written as follows : @xmath277 with the above information , one has enough tools to calculate the matrix elements of the equation ( [ eq25 ] ) .
gribov and l.n .
lipatov , yad .
15 ( 1972 ) 781 .. l.n .
lipatov , sov.j.nucl.phys . ,
20 ( 1975 ) 94 .
g. altarelli and g. parisi , nucl.phys.b , 126 ( 1977 ) 298 .
dokshitzer , sov.phys.jetp , 46 ( 1977 ) 641 .
kimber , a.d .
martin and m.g .
ryskin , phys.rev.d , 63 ( 2001 ) 114027 .
martin , m.g .
ryskin , g. watt , eur.phys.j.c , 66 ( 2010 ) 163 .
kimber , j. kwiecinski , a.d .
martin , a.m. stasto , phys.rev.d 62 , ( 2000 ) 094006 .
kimber , unintegrated parton distributions , ph.d .
thesis , university of durham , u.k .
g. watta , a.d .
martina and m.g .
ryskina , phys.rev.d 70 ( 2004 ) 014012 .
m. ciafaloni , nucl.phys.b , 296 ( 1988 ) 49 . s. catani , f. fiorani , and g. marchesini , phys.lett.b , 234 ( 1990 ) 339 .
s. catani , f. fiorani , and g. marchesini , nucl.phys.b , 336 ( 1990 ) 18 . m. g. marchesini , proceedings of the workshop qcd at 200 tev erice , italy , edited by l. cifarelli and yu.l .
dokshitzer , plenum , new york ( 1992 ) 183 .
g. marchesini , nucl.phys.b , 445 ( 1995 ) 49 .
j. kwiecinski , a.d .
martin , p.j .
sutton , phys.rev.d52:1445 - 1458 ( 1995 ) .
fadin , e.a .
kuraev and l.n .
lipatov , phys.lett.b , 60 ( 1975 ) 50 .
lipatov , sov.j.nucl.phys .
, 23 ( 1976 ) 642 .
kuraev , l.n .
lipatov and v.s .
fadin , sov.phys.jetp , 44 ( 1976 ) 45 .
kuraev , l.n .
lipatov and v.s .
fadin , sov.phys.jetp , 45 ( 1977 ) 199 .
balitsky and l.n .
lipatov , sov.j.nucl.phys . , 28 ( 1978 ) 822 .
m. modarres , h. hosseinkhani , nucl.phys.a , 815 ( 2009 ) 40 . m. modarres , h. hosseinkhani , few - body syst .
, 47 ( 2010 ) 237 .
h. hosseinkhani , m. modarres , phys.lett.b , 694 ( 2011 ) 355 .
h. hosseinkhani , m. modarres , phys.lett.b , 708 ( 2012 ) 75 .
m. modarres , h. hosseinkhani , n. olanj , nucl.phys.a , 902 ( 2013 ) 21 .
m. modarres , h. hosseinkhani and n. olanj , phys.rev.d , 89 ( 2014 ) 034015 . m. modarres , h. hosseinkhani , n. olanj and m.r .
masouminia , eur.phys.j.c , 75 ( 2015 ) 556 .
m. modarres , m.r .
masouminia , h. hosseinkhani , and n. olanj , nucl.phys.a , 945 ( 2016 ) 168185 .
c. albajar et al .
( ua1 collaboration ) , phys.lett.b , 253 ( 1991 ) 503 .
j. alitti et al .
( ua2 collaboration ) , z.phys . , 47 ( 1990 ) 11 .
f. abe et al .
( cdf collaboration ) , phys.rev.lett .
, 76 ( 1996 ) 3070 .
b. affolder et al .
( cdf collaboration ) , phys.rev.lett . , 84 ( 2000 ) 845 .
s. abachi et al .
( d0 collaboration ) , phys.rev.lett .
, 75 ( 1995 ) 1456 .
b. abbott et al .
( d0 collaboration ) , phys.rev.lett .
, 80 ( 1998 ) 5498 .
b. abbott et al .
( d0 collaboration ) , phys.rev.d , 61 ( 2000 ) 072001 .
b. abbott et al .
( d0 collaboration ) , phys.rev.d , 61 ( 2000 ) 032004 .
b. abbott et al .
( d0 collaboration ) , phys.lett.b , 513 ( 2001 ) 292 .
s. p. baranov , a.v .
lipatov , and n. p. zotov , phys.rev.d , 78 ( 2008 ) 014025 .
m. deak , transversal momentum of the electroweak gauge boson and forward jets in high energy @xmath278 factorisation at the @xmath22 , ph.d .
thesis , university of hamburg , germany , 2009 .
martin , w.j .
stirling , r.s .
thorne , g. watt , eur.phys.j.c , 63 ( 2009 ) 189 .
martin , w.j .
stirling , r.s .
thorne , g. watt , eur.phys.j.c , 64 ( 2009 ) 653 .
martin , w.j .
stirling , r.s .
thorne , g. watt , eur.phys.j.c , 70 ( 2010 ) 51 .
l. a. harland - lang , a. d. martin , p. motylinski , r.s .
thorne , eur.phys.j.c , 75 ( 2015 ) 204 .
j. kwiecinski , a.d .
martin and a.m. stasto , phys.rev.d , 56 ( 1997 ) 3991 .
k. golec - biernat and a.m. stasto , phys.rev.d , 80 ( 2009 ) 014006 .
g. watt , parton distributions , ph.d .
thesis , university of durham , u.k .
kimber , a.d .
martin and m.g .
ryskin , eur .
phys.j.c , 12 ( 2000 ) 655 .
w. furmanski , r. petronzio , phys.lett.b , 97 ( 1980 ) 437 .
baranov , a.v .
lipatov and n.p .
zotov , phys.rev.d , 81 ( 2010 ) 094034 .
lipatov and n.p .
zotov , phys.rev.d , 81 ( 2010 ) 094027 ; phys.rev.d , 72 ( 2005 ) 054002 .
vermaseren , symbolic manipulation with form , published by computer algebra , nederland , kruislaan 413 , 1098 , sj amsterdaam , 1991 ; isbn 90 - 74116 - 01 - 9 .
m. cacciari , g. p. salam and g. soyez , jhep 0804 ( 2008 ) 063 .
yao et al . ( particle data group ) , j. phys .
g 33 , 1 ( 2006 ) .
p. jimenez - delgado , e. reya , phys.rev.d , 79 ( 2009 ) 074023 .
m. glck , e. reya , mod .
phys.lett.a , 22 ( 2007 ) 351 .
p. jimenez - delgado , e. reya , phys.rev.d , 80 ( 2009 ) 114011 .
atlas collaboration ( georges aad ( marseille , cppm ) et al . ) , eur .
j. c 76(5 ) ( 2016 ) 1 - 61 .
cms collaboration ( vardan khachatryan ( yerevan phys .
et al . ) , hys .
b 749 ( 2015 ) 187 . a. gehrmann - de ridder , t. gehrmann , e.w.n .
glover , a. huss , t.a .
morgan , ( 2016 ) [ arxiv:1605.04295 [ hep - ph ] ] . and @xmath64 in the @xmath8-factorization framework .
the @xmath279 and @xmath280 represent the corresponding updf in the kmr , the lo - mrw or the nlo - mrw frameworks , i.e. the equations ( [ eq6 ] ) , ( [ eq7 ] ) , ( [ eq9 ] ) and ( [ eq11 ] ) . ]
sub - process , panel ( b ) to the @xmath281 sub - process and panel ( c ) to the @xmath282 sub - process .
it should be pointed out that one may find additional non - ladder - type diagrams which contribute to these matrix elements .
we have eliminated these un - desirable contributions using our choice of the gluon gauge , the equation ( [ eq21 ] ) . ]
bosons in a dis at @xmath283 , against the transverse momentum distribution of the produced particle .
the panels ( a ) and ( b ) illustrate the up - down and charm - strange contributions , respectively .
the contribution of each partonic sub - process is singled out : the green - dash histogram is for @xmath284 , the red - dotted histogram is for @xmath285 and the blue - dash - dotted histogram is for @xmath286 .
the black - full histogram is the total contribution of the give quark pairs .
the histograms are produced using the @xmath3 @xmath7 with the @xmath13 of @xmath87 .
] boson in a dis at @xmath283 , against the transverse momentum distribution of the produced particle .
the contributions of the up and the down quarks ( the panels ( a ) and ( b ) , respectively ) and the lightest sea - quarks ( the panel ( c ) for the charm quark and the panel ( d ) for the strange quark ) . the green - dash histogram is for @xmath287 , the red - dotted histogram is for @xmath288 and the blue - dash - dotted histogram is for @xmath289 .
the black - full histogram is the total contribution of the given quark .
the data is produced using the @xmath3 @xmath7 , with the @xmath13 of @xmath87 . ]
bosons in a dis at @xmath283 , versus the transverse momentum distribution of the produced particle .
the panels ( a ) , ( b ) and ( c ) correspond to the @xmath290 , @xmath291 and @xmath292 sub - processes , respectively .
the data have been obtained using the @xmath7 of @xmath3 , with the @xmath13 of @xmath87 . ]
production in the @xmath11 in the @xmath3 ( the panel ( a ) ) , @xmath9 ( the panel ( b ) ) and @xmath10 ( the panel ( c ) ) frameworks .
the panel ( d ) illustrates this comparison with the help of experimental data of @xmath71 collaboration , the reference @xcite .
the panels ( e),(f ) and ( g ) are the same values , but this time they deviled by the total cross - sections in their respective framework and compared to an older set of data points from the @xmath71 collaboration , @xcite .
again , an overall comparison with the experiment is presented in the panel ( h ) . to perform these calculations ,
we have utilized the @xmath13 of @xmath87.,title="fig : " ] production in the @xmath11 in the @xmath3 ( the panel ( a ) ) , @xmath9 ( the panel ( b ) ) and @xmath10 ( the panel ( c ) ) frameworks .
the panel ( d ) illustrates this comparison with the help of experimental data of @xmath71 collaboration , the reference @xcite .
the panels ( e),(f ) and ( g ) are the same values , but this time they deviled by the total cross - sections in their respective framework and compared to an older set of data points from the @xmath71 collaboration , @xcite .
again , an overall comparison with the experiment is presented in the panel ( h ) . to perform these calculations ,
we have utilized the @xmath13 of @xmath87.,title="fig : " ] production in the @xmath11 in the @xmath3 ( the panel ( a ) ) , @xmath9 ( the panel ( b ) ) and @xmath10 ( the panel ( c ) ) frameworks .
the panel ( d ) illustrates this comparison with the help of the experimental data of @xmath71 and @xmath72 collaborations , the references @xcite . ]
boson in @xmath293 .
the labels ( a ) , ( b ) and ( c ) compare the contributions of the individual sub - processes in their respective frameworks .
the total values of differential cross - section in these frameworks are subjected to a comparison with the data of the @xmath71 collaboration @xcite separately , in the label ( d ) .
this very same notion is also presented in the labels ( e ) through ( f ) , where the @xmath294 histograms are being compared with each other and with the data from @xcite.,title="fig : " ] boson in @xmath293 .
the labels ( a ) , ( b ) and ( c ) compare the contributions of the individual sub - processes in their respective frameworks .
the total values of differential cross - section in these frameworks are subjected to a comparison with the data of the @xmath71 collaboration @xcite separately , in the label ( d ) .
this very same notion is also presented in the labels ( e ) through ( f ) , where the @xmath294 histograms are being compared with each other and with the data from @xcite.,title="fig : " ] production , using the @xmath7 of @xmath3 , prepared with the @xmath13 of @xmath87 ( label ( a ) ) and @xmath88 ( label ( b ) ) .
label ( c ) shows their difference relative to the experimental data of the @xmath71 collaboration , reference @xcite . ]
production , using the @xmath7 of @xmath3 , prepared with the @xmath13 of @xmath87 ( label ( a ) ) and @xmath88 ( label ( b ) ) . label ( c ) shows their difference relative to the experimental data of the @xmath71 and @xmath72 collaborations , references @xcite . ] , calculated in different frameworks , against the transverse momentum of the produced gauge boson at @xmath296 .
the notions of the histograms are as follows : the continues black histogram represents the calculation in using the @xmath3 @xmath7 , the dotted green histogram is prepared in @xmath21 @xmath5 framework and the short - dotted red in the @xmath11 @xmath5 . to perform these calculations ,
we have utilized the @xmath13 of @xmath87 .
the brown dot - dot - dashed histogram in produced using the @xmath47 @xmath42 ( reference @xcite ) .
the yellow dotted - dashed histogram is calculated , utilizing the doubly unintegrated parton distributions ( dupdf ) in the framework of @xmath227-factorization , reference @xcite .
the purple short - dashed histogram is calculated in the collinear framework . ]
bosons as a function of the center - of - mass energy , @xmath298 .
the experimental data are acquired from the @xmath299 , @xmath300 , @xmath71 and @xmath72 collaborations , references @xcite .
the calculations are performed using the @xmath3 and the @xmath198 @xmath7 .
we have omitted the @xmath11 @xmath7 results here , to save computation data . ]
boson in @xmath301 , using the @xmath3 approach .
the individual contributions from the partonic sub - processes are presented and the total values of ( single and double ) differential cross - sections are subjected to comparison with the data of the @xmath231 ( black circles ) and @xmath232 ( white circles ) collaborations @xcite .
the labels ( a ) through ( f ) illustrate the results of our calculations for single differential cross - section of the production of @xmath64 , in the given rapidity regions .
the results for double differential cross - section are presented in the this figure with labels ( g ) through ( h).,title="fig : " ] boson in @xmath301 , using the @xmath3 approach .
the individual contributions from the partonic sub - processes are presented and the total values of ( single and double ) differential cross - sections are subjected to comparison with the data of the @xmath231 ( black circles ) and @xmath232 ( white circles ) collaborations @xcite .
the labels ( a ) through ( f ) illustrate the results of our calculations for single differential cross - section of the production of @xmath64 , in the given rapidity regions .
the results for double differential cross - section are presented in the this figure with labels ( g ) through ( h).,title="fig : " ] | in a series of papers , we have investigated the compatibility of the @xmath0-@xmath1-@xmath2 ( @xmath3 ) and @xmath1-@xmath2-@xmath4 ( @xmath5 ) @xmath6 parton distribution functions ( @xmath7 ) as well as the description of the experimental data on the proton structure functions .
the present work is a sequel to that survey , via calculation of the transverse momentum distribution of the electro - weak gauge vector bosons in the @xmath8-factorization scheme , by the means of the @xmath3 , the @xmath9 and the @xmath10 @xmath7 , in the next - to leading order ( @xmath11 ) . to this end , we have calculated and aggregated the invariant amplitudes of the corresponding @xmath12 diagrams in the @xmath11 , and counted the individual contributions in different frameworks .
the preparation process for the @xmath7 utilizes the @xmath13 of @xmath1 et al , @xmath14 , @xmath15 , @xmath16 and @xmath17 as the inputs .
afterwards , the results have been analyzed against each other , as well as the existing experimental data .
our calculation show excellent agreement with the experiment data .
it is however interesting to point - out that , the calculation using the @xmath3 framework illustrates a stronger agreement with the experimental data , despite the fact that the @xmath9 and the @xmath10 formalisms employ a better theoretical description of the @xmath18 evolution equation .
this is of course due to the use of the different implementation of the angular ordering constraint in the @xmath3 approach , in which automatically includes the re - summation of @xmath19 , @xmath20 logarithms , in the @xmath21-@xmath18 evolution equation
. 1 cm |
the development of new algorithms and of powerful computers currently allows to study in explicit solvent large conformational changes of small proteins and peptides@xcite and smaller conformational changes in large proteins@xcite , but still with some computational effort .
more complex simulations , like those of large changes in large systems , of aggregation of many protein chains , or of systematic mutation scans still require the use of models with simplified degrees of freedom .
pasrticularly useful in this respect are implicit - solvent models controlled by simple potentials , like those involving only contact functions ( and thus not requiring the lengthy calculation of accessible surface areas ) . while a thorough sampling of the conformational space of a protein system described by such simplified models is now rather affordable even for large proteins and even describing explicitely all the heavy atoms of the system , the determination of a simple potential capable of recapitulate the properties of a protein is still a challenging problem .
the basic requirement for such a potential is to make the native conformation of proteins stable , as entailed by the thermodynamic hypotesis @xcite .
several different approaches were used to implement this requirement . using associative - memory potentials
@xcite which encodes for correlation between protein sequence and native structure , motivated by the theory of neural networks , it was possible to predict the native conformation of a number of proteins from the knowledge of their sequence @xcite , even if in a simplified geometry . minimizing simulateneously the potential in the native conformation of several proteins with respect to their competitive conformations @xcite allowed to design a potential capable of identifying the native conformation within the framework of a minimal model of protein - like polymers .
however , this potential failed to distinguish the native from alternative conformations in the case of real proteins , mainly because the exploration of competitive conformations was computationally too demanding .
similar approaches were carried out sampling competitive conformations with a monte carlo algorithm within a bead model @xcite , or through a variational approach@xcite .
anyway , they have not been completely succesful for real proteins , always yielding poor results for full - sized protein molecules .
simple potentials originally designed for structure prediction have also been succesfully used to sample non - native states of small proteins @xcite .
a simpler approach is to use structure - based models , specific for each protein . in this case one
desists from building a universal potential , capable of predicting the native conformation from its sequence , and focuses on the investigation of the properties of a protein of known structure .
this is the case , for example , of the popular go model @xcite , which is a direct implementation of the principle of minimal frustration @xcite .
for example , with a coarse - grained go model it was possible to simulate the cotranslational folding of 100-residues proteins within the whole , explicitly - represented ribosome @xcite .
structure - based models have been succesful in reproducing a number of features of proteins , expecially related to the native state and to the transition between the native and the denatured state @xcite . however , they are not able to describe properly non - native interactions , and consequently can not account for the properties of the denatured state , for intermediate states stabilized by non - native interaction , for protein aggregation , and for all those properties that emerge from the competition between native and non - native interactions . in the present work we build an implicit - solvent , model which describes all heavy atoms and which retains the computational handiness of minimally - frustrated models , but do not suffer their limitations .
a key feature of this model is that it must fold to the native conformation of the protein . for this purpose
, we develop a strategy to design a potential between the different atom types to make the equilibrium state of the model at low temperature unique and equal to the native conformation .
once the potential has been designed , one can sample the conformational space of the system , and thus study thermodynamic quantities other than those used as input in the design algorithm . in this way it is possible to understand what properties of the protein are a necessary consequence of the stability of its native state .
we show that the low - temperature equilibrium state is unique and is identical to the experimental the two - state character of the transition between native and denatured state , the energetic effect of experimentally - characterized mutations and some features of the denatured state can be reproduced without any further input to the system .
the potential is chosen as the sum of two - body terms , accounting for the interactions between pairs of atoms , and shaped as a double spherical well .
this choice allows a remarkably fast sampling of the conformational space of the protein system by means of monte carlo ( mc ) algorithms .
each two - body term is determined by a single parameter which determines the depth of the energy wells and which depends of the chemical species involved .
operatively , the set of energy parameters associated with all pairs of chemical species are optimized according to an iterative monte carlo algorithm , employing the reweighting scheme developed by norgaard and coworkers @xcite , to make the thermal averages of a set of inter - atomic distances match the value they display in the experimental native conformation of the protein
. a sequence - dependent potential on the backbone dihedrals is also introduced to favour the formation of secondary structures
. the resulting potential will be minimally frustrated if this is required by the system to display a stable native state , but this ingredient is not pushed by hand . in fact , atoms of the same type but belonging to different positions along the chain interact in the same way , and consequently can stabilize , even strongly , non - native interactions .
in the model we developed proteins are described through all their heavy atoms .
all bond distances , backbone angles and dihedrals of the peptidic bond are mantained rigidly fixed , corresponding to their experimental values .
the ramachandran dihedrals can move freely , while the residue can move among the rotamers defined in ref .
@xcite .
starting from the knowledge of the native conformation of the protein , the potential which controls it has the form @xmath0 the former , two body term is a two - well spherical potential which depends on the positions @xmath1 and on the kind @xmath2 of the atoms involved , in the form @xmath3 atoms that are separated by less than 9 other atoms along the backbone do not display attractive two body interactions .
the minimum of the well has energy @xmath4 that depends on the types @xmath5,@xmath6 of the atoms involved .
different atoms in different amino acids are regarded as different atom types , giving a total of 163 atom types . defining a native contact between two atoms if the two atoms are closer than @xmath7 , we label @xmath8 the maximum distance between atoms of kind @xmath5 and @xmath6 in all native contacts of the protein . the hard - core radius for that pair of atom types is then defined as @xmath9 , at the radius @xmath10 the energy depth of the well is decreased by a factor 2 , and the overall interaction range is @xmath11 .
the potential on the ranachandran dihedrals @xmath12 and @xmath13 is meant to account for the interactions between atoms close along the chain , and thus to induce the formation of local secondary structure .
it has the form @xmath14,\end{aligned}\ ] ] where @xmath15 are the energy constants that set the weight of the dihedral potential with respect to the two body potential and to each other and are chosen as @xmath16 and @xmath17 in order to allow the formation of secondary structure at @xmath18 but , at the same time , not to make the two body potential irrelevant@xcite .
the quantities @xmath19 , @xmath20 , @xmath21 and @xmath22 are the averages of ramachandran dihedrals in typical @xmath23 and @xmath24 conformations , respectively , while the quantities @xmath25 , @xmath26 , @xmath27 and @xmath28 are the associated standard deviations ( see @xcite ) .
the quantities @xmath29 and @xmath30 are the sequence
dependent propensities for the @xmath31th amino acid of @xmath23 and @xmath24 structure , respectively , calculated with psipred@xcite .
we choose not to make it dependent on the specific native conformation not to bias the formation of secondary structures which could be stabilized by tertiary contacts .
the dihedral potential is not affected by the optimization procedure . before starting the simulation , for each protein a set of @xmath32 pairs of atoms
@xmath33 are selected in such a way that they do not belong to amino acids closer than 4 along the sequence , and the distances @xmath34 between each pair in the native conformation recorded .
this choice guarantees that the implementation of all the @xmath34 in a conformation of the protein makes it identical to the native conformation , with an rmsd smaller than 1 .
the whole idea is to optimize the interaction matrix @xmath4 so that the thermal average of the distance between each pair of atoms @xmath35 and @xmath36 is equal to the distance @xmath34 they have in the experimental native conformation , that is @xmath37 and consequently that the equilibrium conformation of the protein is the native one . to implement this idea , we start from an interaction matrix in which @xmath38 if there is al least one pair of atoms @xmath5 and @xmath6 such that @xmath39 and 0 otherwise .
the choice of the initial matrix is not really critical .
making use of the potential ( [ eq : potential ] ) , a mc sampling is carried out and a set of conformations at temperature @xmath40 , which is regarded as reference temperature and sets the energy units ( boltzmann s constant is also set to 1 ) , is recorded . at the end of the mc sampling , the average distances @xmath41 are calculated from the recorded conformations and the @xmath42 between them and the native distances @xmath34 is evaluated , using @xmath43as error allowed for all contacts in the definition of @xmath42 .
the @xmath44 are optimized to minimize the @xmath42 making use of a zero - temperature random minimization . at each step of the minimization ,
the average distances @xmath45 according to the modified potential @xmath46 are calculated following the reweigting scheme described in ref .
@xcite , that is @xmath47,\ ] ] where @xmath48\ ] ] and the index @xmath49 runs over 5000 conformations recorded during the mc sampling carried out with the potential @xmath50 .
then , a new mc simulation is carried out with the new potentials and the procedure is repeated iteratively 100 times .
the mc sampling is carried out with a parallel - tempering @xcite scheme .
the mc moves are pivots on the backbone dihedrals , combinations of pivots on adjacent backbone dihedrals @xcite to produce local moves , and discrete moves of the side chains among all possible rotamers . in each simulation 8
replicas of the system are used , at temperatures ranging from 1 to 1.75 .
each mc iteration is carried out for @xmath51 steps for each replica .
every @xmath52 steps after the half of the simulation the conformation belonging to the replica at @xmath40 is recorded .
more details about the model and the optimization scheme are given in @xcite .
important questions concerning the optimization procedure are whether the optimal potential is unique and to which extent it is portable among different proteins .
a comparison of two interaction matrices for protein g , optimized independently on each other , give a correlation coefficient of 0.74 , with matrix elements more similar towards the ends of the distribution and more dissimilar towards zero@xcite .
this suggests that the most stabilizing matrix elements are rather independent on the realization of the optimization procedure , but depends only on the protein . on the other hand , the correlation between the matrix elements associated with the same atom types in two proteins , specifically protein g and villin@xcite , is 0.08 , indicating that the optimized potential is not portable among proteins .
a necessary condition that the optimized models have to satisfy is to display the experimental native conformation as low temperature equilibrium state .
although the optimization was carried out towards the native distances , it is not straightforward that this is enough to let the model satisfy such a necessary condition . in the present model the interaction between two atoms depend on their kind , not on their position in the protein
this introduces frustration@xcite in the system as , differently from the go models@xcite , the optimal interaction matrix is not simply that in which two atoms strongly attract each other if they are in contact in the experimentally - determined native conformation .
the model satisfies the above necessary conditions if the optimization procedure is able to lower the energy of the native conformation below that of the competing conformations@xcite or , in other words , if it can minimize its degree of frustration@xcite .
we have tested the model on three widely studied proteins .
these are the villin headpiece ( pdb code 1vii ) , the b1 domain of protein g ( pdb code 1pgb ) and the sh3 domain of src ( pdb code 1fmk ) .
the optimization procedure is illustrated in fig .
[ fig : chi2 ] , where the @xmath42 to the set of native distances and the average rmsd to the native conformation is displayed as a function of the number of iterations .
each iteration consists of a mc sampling and an optimization of the interaction matrix . in the case of protein g and sh3
there is a sharp drop of both @xmath42 and average rmsd in the first 20 iterations .
protein g reaches a stationary @xmath53 and an average rmsd @xmath54 nm , while sh3 reaches @xmath55 and an average rmsd @xmath54 nm .
interestingly , while , the average rmsd reaches it stationary value around the 20th iteration and remains stationary since then , the @xmath42 takes a longer time to find its minimum , indicating that rmsd does not capture completely all structural features of the native state .
the behavior of villin is more noisy , most likely because its size is smaller than that of the other two proteins .
anyway , it can converge to @xmath55 and @xmath56 nm after 100 iterations .
the minimum energy conformations found with the interaction matrix obtained in the last iteration of the optimization process is displayed in figure [ fig : native ] for each of the three proteins .
the rmsd to the experimental native conformations are @xmath57 nm for villin , @xmath58 nm for protein g and @xmath59 nm for sh3 .
no low energy conformations with rmsd markedly larger than these are observed@xcite .
the thermodynamic properties of the three proteins as a function of temperature , calculated with a weighted histogram algorithm @xcite , are summarized in figs .
[ fig : cv_villin ] , [ fig : cv_protg ] and [ fig : cv_sh3 ] , respectively .
all of them display two peaks in the specific heat .
the lower temperature one ( at @xmath60 for villin , @xmath61 for protein g and @xmath62 for sh3 ) marks the folding transition , as testified by the change and in average rmsd ( calculated on all heavy atoms ) and fraction of native contacts @xmath63 that takes place at those temperatures .
the higher temperature peak ( @xmath64 for villin , @xmath65 for protein g and @xmath66 for sh3 ) corresponds to the coil
globule transition ( cf .
the change in average gyration radius at those temperatures ) . in agreement with the experimental findings , and
not unexpectedly because of their difference in size , villin results less stable than protein g ( folding temperatures at neutral ph are @xmath67c for villin@xcite and @xmath68c for protein g@xcite ) , and the folding transition less cooperative . in fact , the ratio @xmath69 between calorimetric and vant hoff enthalpy , which takes its minimum value of 1 for a pure two state transition@xcite , results from model calculations to be @xmath70 for villin and @xmath71 for protein g , to be compared with the experimental values @xmath72 for villin@xcite and @xmath73 for protein g@xcite , while it is @xmath74 for our model of sh3 .
however , in all cases the model understimate the two body character of the folding transition , as already observed for other models which only include two body interactions@xcite .
it should be noted that the model displays a folding transition for all the three proteins at temperatures lower than 1 , that is the temperature at which the interaction potential has been optimized to reproduce the native distances .
this suggests that the computational limitations in the optimization of the interaction matrix result not much in errors in the conformational properties of low temperature states , but in a decreased thermodynamic stability .
the free energies of the three proteins as a function of the rmsd and of the gyration radius , calculated with a weighted histogram algorithm @xcite , are displayed in fig .
[ fig : free ] in the case of a temperature below the folding transition , a temperature between the folding and the coil globule transition and a temperature above the coil globule transition . for none of the proteins
the free energy profile highlights detectable intermediates .
the globular denatured state ( at @xmath75 ) is in all cases rather native like , displaying rmsd of the order of 0.50.6 nm . the reason for such a low rmsd is the formation of residual , largely native like , structure in the denaturated state , as shown in fig .
[ fig : stride ] . in the case of villin , residual alpha
helical structure is larger in the n terminal segment , slightly smaller in the c - terminal segment , and marginal in the central segment .
these ratios are in agreement with circular dochroism spectra of isolated fragments of villin@xcite and with explicit solvent molecular dynamics simulations@xcite . the denatured state of protein g displays in native like residual structure in the two hairpins and in the helix , but not the non native turns osberved in the acid denatured state by nmr@xcite .
the denaturated state of sh3 is enriched in beta - starnd structure , a feature that is not observed in nmr experiments with urea , which indicates abundance of non - native helices@xcite
. there can be two straightforward reasons for this dicrepancy .
first , our model simulates a thermal denatured state , while in nmr experiments the protein is destabilized by urea .
moreover , while the agreement with experiments of the other two proteins concerns native like structure , in the case of sh3 the model is not able to predict non native residual structure .
this could be due to the fact that the optimization of the potential to stabilize the native conformation over minimize the frustration of the system .
however , a nice feature of the present approach is that , in principle , one can optimize the interaction matrix to reproduce the native distances at low temperature and , simultaneously , the data observed in the denatured state at higher temperature .
in the case of protein g and sh3 , the free energy changes @xmath76 of the native state upon mutation was measured for a large number of mutations@xcite . within the present model
, the relatively small computational cost of sampling the conformational space allows to simulate the effect of each mutation , and compare the result with the experimental data .
these simulations have two goals .
first , the comparison between experimental and calculated @xmath76 can contribute to validate the model .
moreover , the simulation has access to conformational properties of the mutated system that can not be studied experimentally in a direct way .
operatively , a mutation means changing the atom types of the mutated residue , which interact with the same matrix elements of the wild type protein ( no further optimization is carried out ) , and updating the secondary structure propensities in the dihedral potential .
for each of the mutation reported for protein g@xcite and sh3@xcite we have carried out an equilibrium simulation , obtaining the free energy profile of the wild type ( @xmath77 ) and of the mutated ( @xmath78 ) protein , as a function of rmsd and exposed area@xcite @xmath79 of the tryptophanes .
the reason for the choice of @xmath79 is that experimental @xmath76 were obtained from kinetic experiments in which the measured quantity is the fluorescence of the tryptophanes , which depend on their molecular environment . from these free energies
, we have calculated the free energy differences in a two state approximation , that is @xmath80 where @xmath81 \nonumber\\ p_n^{mut}&\equiv\int_{\cal n}d\,\text{rmsd}\,da_w\;\exp[-f^{mut}(\text{rmsd},a_w)/t ] \end{aligned}\ ] ] and the native region @xmath82 in the free energy profiles
is that defined in fig .
[ fig : freemut ] .
the comparison between experimental and computed @xmath76 is displayed in figs .
[ fig : mut_g ] and [ fig : mut_sh3 ] for protein g and sh3 , respectively .
the correlation coefficients are , respectively , 0.57 and 0.50 , which increase , respectively , to 0.79 and 0.73 if we exclude four outliers .
these values correspond to the optimal choice of the native region @xmath82 .
interestingly , such outliers correspond to sites which display in the calculations large native like structure or does not display the non native secondary structures measured by nmr@xcite .
consequently , one could make the hypothesis that the poor agreement between theoretical and experimental @xmath76 is associated with the overstimation of native structure in the denatured state discussed in the previous section .
moreover , it is interesting to note that the good overall correlation with the experimental data can be obtained only defining the native state using rmsd and @xmath79 . calculating the values of @xmath83 and @xmath84 as integral over rmsd and gyration radius ( cf .
[ fig : free ] ) , on rmsd only , or on @xmath79 only give correlations in the range 0.20.4 .
the reason for this difference in the results seems to be that rmsd and @xmath79 are less correlated than rmsd and @xmath85 ( cf . figs .
[ fig : free ] and [ fig : freemut ] ) , and consequently are better in defining the native region . specifically , the effect of mutations increase the probability of conformations with values of @xmath79 larger than that of the wild type protein , mantaining a rather small rmsd , as shown in fig . [
fig : freemut ] .
a nice feature of the potential developed above is that , being defined with respect to atom types ( and not on atom identifiers , like in go models ) , it include some degree of frustration , which is known to be present in proteins@xcite .
one can thus inspect the energy map of the native conformations of the three proteins already discussed , to identify repulsive contacts , defined as those displaying @xmath86 .
such contacts are highlighted in red in fig .
[ fig : frust ] .
there are 11.9% frustrated contacts in villin , 8.6% in protein g and 20.3% in sh3 , numbers that are comparable to those found in similar calculations carried out with other potentials@xcite . in the case of villin
, they are localized mainly in the third helix and in the tertiary contacts between the first helix and the other two .
this agrees qualitatively with the result of a similar investigations carried out with the help of an evolutionary
derived potential@xcite and of an associative
memory potential@xcite , which emphasise the frustration of contacts within the third helix and between the first and the second helix . in the case of protein g ,
the present model identifies frustrated contacts in the helix and in the terminal part of the first hairpin , while the associative memory potential in the helix and in the secon hairpin . in the case of sh3
, the optimized potential reveals frustrated contacts in the terminal beta - sheet , between the rt loop and the distal hairpin and in the stem of the distal hairpin , while the associative memory potential in the stem of distal hairpin and in the stem of of the rt loop and the evolutionary derived potential in the stem of the distal hairpin , in the rt loop , between these two and in the n src loop .
the small differences observed in the frustration maps generated in present and in other works are most probably due to the fact that our potential is atom - based , while the others are amino acid - based .
this means that repulsive and attractive interaction between pairs of atoms between two given amino acids , as predicted by the present model , can sum together to give a total interaction which can be either repulsive or attractive .
consequently , the present model provide an information which is complementary to that of the other two .
in spite of the continuously growing capability of algorithms and computers to perform longer simulations of larger systems with portable , explicit solvent potentials , implicit solvent models of biomolecules interacting with simplified potentials can still be useful for many applications , like very - large systems , mutation scans and aggregation studies .
so far , this kind of problems were tackled making use of go models , which neglects the residual frustration present in all proteins . the model discussed in the present work
is based on an optimization of the matrix which controls the interaction between atom types to make the experimental native conformation as the low temperature equilibrium state of the system .
this model can reproduce a number of known data about proteins , like the stability of their native state , the two state transition , the energetic effect of mutations on their stability , while still displaying a realistic degree of frustration .
we think that the strength of this approach is its versatility .
one can use as input for the optimization of the potential any set of experimental data , even an heterogeneous one , provided that they can be expressed as thermal averages of some conformational property .
for example , we showed that the structure of the denatured state of the proteins used in the present work is not in complete agreement with the nmr data in denaturing conditions .
this is not really unexpected , since the input data we used describe the native conformation , and consequently the predictions of the model can not but worsen as they involve states which are distant from the native state . to improve the model
, one can thus introduce in the optimization data concerning the denatured state .
moreover , this approach can be used to correct existing potentials , even in explicit solvent , for specific goals .
it is enough to use the potential to be corrected as initial potential of the optimization procedure .
99 s. piana , k. lindorff - larsen and d. e. shaw , j. phys .
chem b * 117 * , 12935 ( 2013 ) .
g. r. bowman , v. a. voelz and v. s. pande , j. am .
* 133 * , 664 ( 2011 ) .
l. sutto and f. l. gervasio , proc .
usa * 110 * , 10616 ( 2013 ) . c. b. anfinsen , science , * 181 * , 223 ( 1973 ) .
r. a. goldstein , z. a. luthey schulten and p. g. wolynes , proc .
usa * 89 * , 4918 ( 1992 ) m. c. prentiss , c hardin , m. p. eastwood , c. zong and p. g. wolynes , j. chem .
theo . comp . * 2 * , 705 ( 2006 ) .
l. a. mirny and e. i. shakhnovich , j. mol
. biol . * 264 * , 1164 ( 1996 ) .
m. h. hao and h. a. scheraga , proc .
usa , * 93 * , 4984 ( 1996 ) .
f. seno , and a. maritan , proteins struct .
. gen . * 30 * , 244 ( 1998 ) .
f. seno , c. micheletti and a. maritan , a. phys . rev
81 * , 2172 ( 1998 ) .
d. shirvanyants , f. ding , d. tsao , s. ramachandran and n. v. dokholyan , j. phys .
b * 116 * 8372 ( 2012 ) .
s. kimura , m. caldarini , r. a. broglia , n. v. dokholyan and g. tiana , proteins struct .
( in press ) n. go , annu .
. bioengin .
* 12 * , 183 ( 1983 ) .
j. d. bryngelson and p. g. wolynes , proc .
usa * 84 * , 7524 ( 1987 ) .
a. h. elcock , plos comp . biol . * 2 * , e98 ( 2006 ) . c. clementi , curr .
* 18 * , 10 ( 2008 ) .
r. d. hills and c. l. brooks , int .
sci . * 10 * , 889 ( 2009 ) .
k. wolff , m. vendruscolo and m. porto , phys . rev .
e * 8 * , 041934 ( 2011 ) .
a. b. norgaard , j. ferkinghoff - borg and k. lindorff - larsen , biophys .
j. , * 94 * , 182 ( 2008 ) .
s. c. lovell , j. m. word , j. m. , richardson and d. c. richardson , proteins struct .
. gen . * 40 * , 389 ( 2000 ) .
d. t. jones , j. mol .
biol . * 292 * , 195 ( 1999 ) r. swendsen and j. wang , phys .
rev . lett . * 57 * , 2607 ( 1986 ) .
j. shimada , e. l. kussell and e. i. shakhnovich , j. mol .
, * 308 * , 79 ( 2001 ) . see supplemental material at [ url will be inserted by aip ] for more details about the model and the omptimization scheme .
g. toulouse , commun .
* 2 * , 115 ( 1977 ) .
e. i. shakhnovich and a. m. gutin , proc .
natl . acad .
usa * 90 * , 7195 ( 1993 ) .
a. ferrenberg and r. swendsen , phys .
lett . , * 63 * , 1195 ( 1989 ) .
r. godoy - ruiz , e. r. henry , j. kubelka , j. , hofrichter , v. muoz , j. m. sanchez - ruiz and w. a. eaton , j. phys .
b , * 112 * ) , 5938 ( 2008 ) .
p. alexander , s. fahnestock , t. lee , j. orban and p. bryan , biochemistry * 31 * , 3597 ( 1992 ) p. l. privalov and n. n. khechinashvili , j , mol .
biol . * 86 * , 665 ( 1974 ) .
h. s. chan , proteins struct .
* 40 * , 543 ( 2000 ) .
y. tang , d. j. rigotti , r. fairman and d. p. raleigh , biochemistry * 43 * , 3264 ( 2004 ) . l. wickstrom , a. okur , k. song , v. hornak , d. p. raleigh and c. l. simmerling , j. mol . biol . * 360 * , 1094 ( 2006 ) .
n. sari , p. alexander , p. n. bryan and j. orban , biochemistry * 39 * , 965 ( 2000 ) .
i. rsner and f. m. poulsen , biochemistry , * 49 * , 3246 ( 2010 ) .
e. l. mccallister , e. alm and d. baker , nature struct . biol .
* 7 * , 669 ( 2000 ) .
v. grantcharova , d.riddle , j. santiago and d. baker , d.nature struct . biol . * 5 * , 714 ( 1998 ) .
f. eisenhaber , p. lijnzaad , p. argos , c. sander and m. scharf , j. comput . chem . * 16 * , 273 ( 1995 ) .
m. jenik , r. g. parra , l. g. radusky , a. turjanski , p. g. wolynes and d. u. ferreiro , nucl .
acid res . *
40 * , w348 ( 2012 ) s. lui and g. tiana , j. chem .
phys , * 139 * , 15103 ( 2013 ) between average and native distances ( red curve , in semi log scale ) and the average rmsd to the experimental native conformation ( blue curve ) as a function of the number of iterations of the mc sampling for villin ( upper panel ) , protein g ( middle panel ) and sh3 ( lower panel ) . ] | the current capacity of computers makes it possible to perform simulations of small systems with portable , explicit - solvent potentials achieving high degree of accuracy .
however , simplified models must be employed to exploit the behaviour of large systems or to perform systematic scans of smaller systems . while powerful algorithms are available to facilitate the sampling of the conformational space , successful applications of such models are hindered by the availability of simple enough potentials able to satisfactorily reproduce known properties of the system .
we develop an interatomic potential to account for a number of properties of proteins in a computationally economic way .
the potential is defined within an all - atom , implicit solvent model by contact functions between the different atom types .
the associated numerical values can be optimised by an iterative monte carlo scheme on any available experimental data , provided that they are expressible as thermal averages of some conformational properties .
we test this model on three different proteins , for which we also perform a scan of all possible point mutations with explicit conformational sampling .
the resulting models , optimised solely on a subset of native distances , not only reproduce the native conformations within a few angstroms from the experimental ones , but show the cooperative transition between native and denatured state and correctly predict the measured free
energy changes associated with point mutations .
moreover , differently from other structure - based models , our method leaves a residual degree of frustration , which is known to be present in protein molecules . |
it has long been known that the kinematics of spiral galaxies do not show keplerian fall - off in their rotation curves , but rather imply the presence of an invisible mass component in addition to the stellar and gaseous disks @xcite . in the local universe , observing the distribution of star light and mapping the gaseous component through hi it has been possible to build up a picture of the how the baryonic component of disk galaxies is distributed and how this relates to the underlying dark matter component ( e.g. @xcite ) .
tracing the evolution of galaxy mass from high - redshift up to the present day is only truly reliable if we can observe the same components at early times .
a pioneering study was performed by @xcite where a handful of rotation curves ( rc s ) of objects at @xmath2 was obtained using traditional longslit spectroscopy on the 10 meter keck telescope . however , at higher redshifts ( e.g. z@xmath31 ) galaxies are much fainter and have smaller angular disk scale lengths than galaxies observed at low redshift , therefore obtaining the spatial information required for detailed studies is beyond the limits of current technology .
indeed , mapping the internal properties and dynamics of both the stellar and gaseous components of galaxies at high redshift is one of the main science drivers for the next generation of ground and spaced based telescopes at many wavelengths ( e.g. elt , ngst , alma ) .
one way to overcome this problem is to use the natural amplification caused by gravitational lensing to boost the size and flux of distant galaxies which serendipitously lie behind massive galaxy clusters .
this technique is extremely useful since we are able to target galaxies which would otherwise be too small and faint to ensure a sufficiently high signal - to - noise spectroscopy in conventional observations . as such
, gravitational lensing has been extensively used to make detailed studies of distant galaxies : for a galaxy at @xmath4=1 with an amplification factor ten , an angular scale of 0.6@xmath5 can correspond to @xmath60.5kpc , sufficient to map also a small spiral .
the benefits of gravitational lensing are complemented by integral field spectroscopy ( which produces a contiguous _ two dimensional velocity _ map at each point in the target galaxy ) .
this allows a clean decoupling of the spatial and spectral information , thus eliminating the problems arising from their mixing in traditional long - slit observations .
it is therefore much easier to identify and study galaxies with regular ( bi - symmetric ) velocity fields . in this paper
, we present a detailed study of four rotation curves extracted from disk galaxies which have been observed through the cores of massive galaxy cluster lenses .
these targets are taken from the recent work of @xcite .
they were observed with the gemini - north multi - object spectrograph integral field unit ( gmos ifu ) .
we concentrate on the galaxy dynamics as traced by the [ oii]@xmath73726.1,3728.8 emission line doublet .
the ifu data provide a map of the galaxy s velocity field in sky co - ordinates . to interpret this field
the magnification and distortion caused by the gravitational lensing effect is removed using detailed models of the cluster lenses ( see @xcite , @xcite , @xcite for details ) .
the primary constraints on the lens models are the positions and redshifts of spectroscopically confirmed gravitational arcs in each cluster .
the source plane velocity fields of four systems which display regular ( bi - symmetric ) rotational velocity fields i.e. resembling rotating disks , were reduced to one - dimensional rotation curves from which the asymptotic terminal velocity was extracted and compared with the galaxy luminosity @xcite .
the key advantage of using gravitational lensing to boost the images of distant galaxies is that we are less biases towards the most luminous galaxies .
whilst observational information on the distribution of the hi disks in galaxies at these early times would be wellcome , such observations will have to wait for future instrumentation ( e.g. alma ) .
in this paper we use nebular emission lines to probe the kinematic of the galaxies .
we extract one - dimensional rotation curves from the velocity fields to infer the distribution of stellar and dark matter components .
finally , we compare our results with similarly luminous disk galaxies in the local universe . through - out
this paper we use a cosmology with @xmath8 , @xmath9 and @xmath10 , @xmath11 .
our sample comes from observations of six gravitational arcs in @xcite . in order to avoid possible biases ,
the targets were selected only to be representative of galaxies in the distant universe and no attempt was made to select galaxies with relaxed late - type morphology .
we did , however , require that arcs were resolved in both spatial dimensions so that a two dimensional velocity field could be extracted from the ifu data .
this restricted our selection to galaxies with moderate magnification . from the sample of six galaxies ,
four galaxies appear to have ( relaxed ) bi - symmetric velocity fields with late - type morphologies and colours .
the rotation curves from these four galaxies appear regular and we therefore restrict our analysis to these arcs . we stress that the morphology , colours and velocity fields of the four galaxies in this sample all strongly suggest these galaxies are consistent with late type spirals ( see @xcite ) . from our optical / near - infrared imaging , we constrain the spectral energy distribution ( sed ) of each galaxy .
since the arcs usually lie with a few arcs - seconds of nearby bright cluster galaxies , we calculate the magnitude of the arcs in various pass - bands by masking the arc and interpolating the light from the nearby cluster members .
the background light is then removed and surface photometry in different bands is obtained . using the cluster mass models the arcs
are reconstructed to the source - plane and the geometry and disk - scale parameters of the disks are measured .
this is achieved by fitting ellipses to an isophote of the galaxy image using a modified version of the idl gauss2dfit routine which fits an exponential profile to the two - dimensional light distribution . from this , the ellipticity , the inclination and luminosities and the disk scale lengths are obtained ( see table 2 and 3 in @xcite ) .
these latter are also reported below in table 1 . in fig .
[ fig : rot_curves ] we show the one dimensional rotation curves of the galaxies in our sample .
these are extracted by sampling the velocity field along the major axis cross section . the zero - point in the velocity
is defined using the center of the galaxy in the reconstructed source plane image .
the error bars for the velocities are derived from the formal @xmath12 uncertainty in the velocity arising from gaussian profile fits to the [ oii ] emission in each averaged pixel of the datacube . for the mass modelling analysis we folded the rotation curves on the kinematical center to ensure that any small - scale kinematical sub - structure is removed
the rotation curves of local spiral galaxies imply the presence of an invisible mass component , in addition to the stellar and gaseous disks .
the paradigm is that the circular velocity field can be characterized by : @xmath13 where the subscripts denote the stellar disk , dark halo and gaseous disk respectively . from the photometry we model the stellar component with a freeman surface density @xcite : @xmath14 where @xmath15 is the disk lenght - scale ,
while @xmath16 can be taken as the `` size '' af the stellar disk , whose contribution to the circular velocity is : @xmath17 where @xmath18 and @xmath19 and @xmath20 are the modified bessel functions computed at @xmath21 . in small spirals
the hi disk is an important baryonic component only for @xmath22 , i.e. outside the region considered here .
it is plausible that in similar objects also at high redshifts , _ inside _ @xmath23 , the h i gas contributes to the gravitating baryonic mass by a very small amount , residing at larger radii and it slowly infalls forming the stellar disk . for the dark matter component we take a spherical halo for which @xmath24 .
following the observational scenario constructed in the local universe @xcite we assume that it has the @xcite density profile ( see also salucci et al 2007 ) : @xmath25 where @xmath26 is the core radius and @xmath27 the effective core density .
it follows that : @xmath28~\eqno(5)\ ] ] with @xmath29 and of course @xmath30 we note that the adopted velocity profile is a quite general : it allows a distribution with a core of size @xmath26 , converges to the nfw profile at large distances and , for suitable values of @xmath26 can also mimic the nfw or a isothermal profile , over the limited region of galaxy which is mapped by the rotation curves .
the mass model has three free parameters : the disk mass @xmath31 , the core radius @xmath26 , and the central core density @xmath27 .
the observations extend out approximately @xmath32 , have 10 - 60 independent measurements with an observational error of @xmath33 in their amplitude , of 0.05 - 0.2 in their slopes @xmath34 .
the error in the estimate of the disk inclination angles is negligible with respect to the above .
these errors are ( understandably ) higher than those associated with the best quality local rc s and make it difficult to constrain the halo density profile @xmath35 . however , they are sufficiently small to yield a reliable value for the halo mass , the average density inside a reference radius , ( which we chose to be @xmath36 so that @xmath37 and the disk mass and a reasonable estimate of the `` core radius '' . by reproducing the observed rotation curves with the models given by equations 1 - 5
we derive the best fit parameters for each galaxy and overlay the resulting mass model onto each rotation curve in fig .
[ fig : rot_curves ] . in table 1
we report the main structural parameters : the disk mass , the halo core radius , @xmath27 , the optical radius @xmath38 and @xmath39 , i.e. the halo contribution to circular velocity at the optical radius . in all
our rotation curves the _ amplitude _ and the _ profile _ of the stellar disk contribution can not reproduce the observed rc rise between 1.5 @xmath40 and the last measured radius .
this strongly suggests evidence for the presence , at @xmath41 , of a dark matter component of mass comparable to that found for local disk galaxies with similar @xmath42 ( see figures 2 , 8 and 9 of persic , salucci & stel , 1996 ) .
we derive disk masses ranging between @xmath43 to @xmath44 for galaxies with reference velocity @xmath42 between 100@xmath45 and 200@xmath45 .
these disk masses are smaller by a factor 2 - 4 than those of the local spirals with the same reference velocity which are shown to follow inner baryon dominance ( see salucci & persic , 1999 ) .
the high redshift stellar disks are sub - maximal disks .
forcing a maximal disk even in the `` weak '' implementation of persic and salucci ( 1990 ) leads to unacceptable fits of our high - z rc s .
the best fit values for @xmath26 is of the order of @xmath46 , which is larger than usually compatible with a nfw profile , although the error - bars on our data preclude any strong statement .
since the gravitational lens model affects the source - plane reconstruction of the galaxies , we must test for the effect that this has on the resulting rotation curves .
we reconstruct the galaxies using the family of lens models that inhabit the @xmath47 contour corresponding to @xmath48 confidence interval relevant to each cluster .
for example , the model of rgb1745 has five free parameters , and the lens modeling uncertainties are therefore derived by considering models within the @xmath49 contour . from each reconstruction
we extract the one - dimensional rotation curve and apply the analysis outlined above and find the maximum variations are : @xmath50 , @xmath51 , @xmath52 .
thus the uncertainties in the gravitational lens modelling is negligable compared to the uncertainties in the rc mass modelling .
a cosmological significance of our result is evident in fig [ fig : md_rhom ] where we compare the disk mass , the mean dark matter density within the optical radius and the angular momentum per unit mass , all as a function of reference velocity .
these are compared to similar properties of the local objects @xcite .
figure [ fig : md_rhom ] strikingly shows that high redshift spirals , modulo an offset of @xmath53 dex , are on the same @xmath54 versus @xmath55 relationship found for local spirals arising from the systematic structural properties of their mass distribution @xcite , see also @xcite . in fig 2 , from the values of @xmath56 , a quantity that differently from @xmath27 is weakly affected by the rc 1-@xmath57 fitting uncertainties , it is apparent that the dm halos of @xmath0 disk galaxies are denser by @xmath58 dex than those around similarly luminous @xmath4=0 spirals .
the evidence that spiral disks at @xmath4=0 and @xmath4=1 have the same structural relationships is further supported by observations of the evolution of the tully - fisher relation ( which correlates the disk mass with @xmath42 ) . in our and in other independent samples ( swinbank et al 2006 , , vogt et al 1996 , bamford et al 2005 ) the galaxies at @xmath0 show a tf relation with a slope similar to that of the local tf , but with an offset compatible with that found in the present work from the disk mass _ vs _ rotation velocity .
this suggests that from @xmath4=1 to @xmath4=0 , the stellar disk masses @xmath59 of a spiral has grown by a factor @xmath60 , that leads to just a modest increase in the dm dominated quantity @xmath42 .
further evidence that @xmath0 disk galaxies are related to present day spirals is provided by the relationship between angular momentum per unit mass ( @xmath61 ) versus the reference velocity , as shown in fig .
[ fig : md_rhom ] . this well theoretically motivated relation ( e.g. @xcite ) can be considered as the imprint of the process of the formation of disks inside dark matter halos related to the cosmological properties of halo spin parameters .
as figure [ fig : md_rhom ] shows , there appears to be no evolution in this crucial relationship between the the comological time at which we observe these spirals , @xmath62 and the present time , @xmath63 .
this agreement is remarkable : it establishes a link between local and high redshift disks , supporting the idea that the angular momentum remains constant during the evolution of a disk system from high redshift to the present day .
the kinematical estimate of the disk mass allows us to derive the mass - to - light ratios for our disk systems as a function of luminosity and colour . in fig .
3 we compare the disk mass and colour as a function of mass - to - light ratio compared to the relation in local spirals @xcite . in order to compare directly with local relations
, we consider a simple passive evolution model for the luminosity evolution . for a single stellar population the zero - point of the local relation
is decreaed by a factor @xmath64 which accounts for the passive evolution of a _ single _ stellar population from @xmath0 to @xmath65 . as fig .
3 shows the mass - to - light ratios as a function of galaxy ( @xmath66 ) colour are in broad agreement with predictions of a single stellar population which is @xmath67 old @xcite , although clearly photometry at other wavelengths ( such as rest - frame k - band ) would allow a more detailed decomposition of the stellar popultations in these galxies .
figure 2 shows that at a radius corresponding to @xmath42 the high redshift galaxies are significantly denser than comparably luminous local disk - galaxies : the average offset is about 0.6 dex in @xmath68 . although we can not exclude that dynamical processes occur between @xmath0 and @xmath65 to reduce the dark - matter density in the luminous regions , this offset is naturally explained if the halos embedding these disk galaxies formed at earlier times than the halos around similarly massive @xmath65 spirals . in this framework
we estimate the ratio between the virialization redshift of the local galaxies and and that of the galaxies in our sample .
since @xmath69 where @xmath70 and @xmath71 are average density and redshift at virialization and @xmath72 is known , for @xmath73 , @xmath74 , which corresponds to @xmath75 . assuming that our sample is a fair representation of disk galaxies at @xmath76 and that these are approximately coeval , from the comparison of their structural properties with those of @xmath65 spirals , the following simple picture emerges : a present day spiral , with a given circular velocity , half - light radius and the angular momentum per unit mass , at redshift @xmath77 had similar values for these quantities , but a smaller stellar mass : @xmath78 . this induces a scale for the average sfr in the past 8 gyr : @xmath79 . with these disks having an average age of 1gyr at @xmath0
we can also derive an `` early times '' average sfr @xmath80 which points towards a declining sfr history .
the marked increase of the luminosity per unit stellar mass in objects at high redshifts with respect to their local counterparts has the simplest explanation in a passive evolution of the starforming disks .
obviously , this simple picture requires us to assume that the high redshift systems are the direct counter - parts of similar rotation speed spirals at low redshift .
in this study , we have investigated the detailed properties of four disk galaxies at @xmath0 .
these galaxies were observed at high spatial resolution thanks to the boost in angular size provided by gravitational lensing by foreground massive galaxy clusters and allow a much more detailed comparison with local populations than usually possible for galaxies at these early times . modelling the one - dimensional rotation curves with those of @xcite we derive
best fit parameters for the total dynamical mass , the core radius , the effective core density and the angular momentum per unit mass .
the best fit model rotation curves to the data show that the amplitude and profile of the stellar disk componentcan not unambiguously reproduce the rise in the rotation curve without a dark matter component . comparing the average dark matter density inside the optical radius
we find that the disk galaxies at @xmath0 have larger densities ( by up to a factor of @xmath1 ) than similar disk galaxies in the local universe . in comparison , we find that the angular momentum per unit mass versus reference velocity is well matched to the local relation suggesting that the angular momentum of the disk remains constant between high redshift and the present day .
though statistically limited , these observations point towards a spirals formation scenario in which stellar disk are slowly grown by the accretion of angular momentum conserving material .
our result , also consistent with the theoretical evolution of the angular momentum of disks from semi - analytic models from @xmath4=1 to @xmath4=0 which show the modest offset of @[email protected] kpc.@xmath45 for objects with circular velocities between 50 and 300@xmath45 ( @xcite ) .
these reults provide an evolutionary link between the disk systems we observe at redshift @xmath821 and the present day population of spirals .
while these results are based on data of only four objects , they nevertheless show the power which gravitational lensing can have on studying the internal properties of high redshift galaxies .
in particular , these observations should be viewed as pathfinder science which will soon be routine with adaptive optics integral field unit observations on eight and ten meter telescopes ( e.g. @xcite ) . moreover , by combining these results with upcoming telescopes and instruments ( e.g. alma ) which will be sensitive enough to map the hi content of galaxies to @xmath0 , the exact relation between gas , stars and dark matter can be probed in much more detail .
we would like to thanks the anonymous referee for his / her suggestions which improved the content and clarity of this paper .
we thank carlton baugh for providing the theoretical evolution of disk mass in galaxies from galform .
we also thank gigi danese for useful discussions .
ams acknowledges support from a pparc fellowship , rgb acknowledges a pparc senior fellowship and irs and gps acknowledges support from the royal society . | we present a detailed study of the structural properties of four gravitationally lensed disk galaxies at @xmath0 . modelling the rotation curves on sub - kpc scales we derive the values for the disk mass , the reference dark matter density and core radius , and the angular momentum per unit mass .
the derived models suggest that the rotation curve profile and amplitude are best fit with a dark matter component similar to those of local spiral galaxies .
the stellar component also has a similar length scale , but with substantially smaller masses than similarly luminous disk galaxies in the local universe . comparing the average dark matter density inside the optical radius
we find that the disk galaxies at @xmath0 have larger densities
( by up to a factor of @xmath1 ) than similar disk galaxies in the local universe .
furthermore , the angular momentum per unit mass versus reference velocity is well matched to the local relation , suggesting that the angular momentum of the disk remains constant between high redshifts and the present day .
though statistically limited , these observations point towards a spirals formation scenario in which stellar disks are slowly grown by the accretion of angular momentum conserving material .
galaxies , rotation curves , galaxies , gravitational lensing , galaxy clusters , integral field spectroscopy , gravitational arcs : individual |
one of the main motivations for r - parity conserving supersymmetry ( susy ) is the prediction that the lightest susy particle ( lsp ) is stable and may be identified as a candidate thermal dark matter ( dm ) particle if it is both electrically neutral and colorless .
quite commonly the role of the lsp is played by the lightest neutralino , @xmath0 , and this be will assumed in the discussion here . while dm searches are superficially focused on the nature of the lsp , the properties of all the other superparticles as well as those of the extended susy higgs sector also come into play .
thus it is impossible to completely separate dm searches from searches for and the examination of the rest of the susy spectrum .
however , even in the simplest susy scenario , the mssm , the number of free parameters ( @xmath1 100 ) is too large to study in all generality .
the traditional approach is to assume the existence of a high - scale theory with only a few parameters ( such as msugra @xcite ) from which all the properties of the sparticles at the tev scale can be determined and studied in detail .
while such an approach is quite valuable @xcite , these scenarios are somewhat phenomenologically limiting and are under increasing tension with a wide range of experimental data including , in some cases , the @xmath2 gev mass of the recently discovered higgs boson@xcite .
one way to circumvent such limitations is to examine the far more general 19-parameter pmssm@xcite .
the pmssm is the most general version of the r - parity conserving mssm when it is subjected to several experimentally - motivated constraints : ( @xmath3 ) cp conservation , ( @xmath4 ) minimal flavor violation at the electroweak scale , ( @xmath5 ) degenerate first and second generation sfermion masses , ( @xmath6 ) negligible yukawa couplings and a - terms for the first two generations .
in particular , no assumptions are made about physics at high scales , e.g. , the nature of susy breaking , in order to capture electroweak scale phenomenology for which a uv - complete theory may not yet exist . imposing the constraints ( @xmath3)-(@xmath6 )
decreases the number of free parameters in the mssm at the tev - scale from 105 to 19 for the case of a neutralino lsp , or 20 when the gravitino mass is included as an additional parameter when it plays the role of the lsp . we have recently begun a detailed study of the signatures for the pmssm at the 7 and 8 ( and eventually 14 ) tev lhc , supplemented by input from dm experiments as well as from precision electroweak and flavor measurements@xcite . in this work ,
we study and compare the sensitivities of present and future experiments employing different approaches to searching for dark matter : collider production , direct and indirect detection .
independently of the lsp type we will not assume that the thermal relic density as calculated for the lsp necessarily saturates the wmap / planck value@xcite to allow for the possibility of multi - component dm .
for example , the axions introduced to solve the strong cp problem may may make up a substantial amount of dm .
the 19 pmssm parameters and the ranges of values employed in our scans are listed in table [ scanranges ] .
the lower and upper limits employed in our scan ranges were chosen to be essentially consistent with tevatron and lep data and to have kinematically accessible sparticles at the lhc , respectively . to study the pmssm
, we generate many millions of model points in this space ( using softsusy@xcite and checking for consistency using suspect@xcite ) , with each point corresponding to a specific set of values for these parameters .
these individual models are then subjected to a large set of collider , flavor , precision measurement , dark matter and theoretical constraints @xcite .
roughly 225k models with a neutralino lsp survive this initial selection and can then be used for further physics studies .
( we have also obtained a model set of similar size for the case of gravitino lsps . ) decay patterns of the susy partners and the extended higgs sector are calculated using a modified version of susy - hit@xcite .
in addition to these two large pmssm model sets , we have recently generated a smaller , specialized neutralino lsp set of @xmath1 10k ` natural ' models all of which predict @xmath7 gev , have an lsp that _ does _ saturate the wmap relic density and which produce values of fine - tuning ( ft ) better than @xmath8 using the barbieri - giudice measure @xcite .
this low - ft model set will also be used in the future as part of our completed study .
.scan ranges for the 19 ( 20 ) parameters of the pmssm with a neutralino ( gravitino ) lsp .
the gravitino mass is scanned with a log prior .
all other parameters are scanned with flat priors , though we expect this choice to have little qualitative impact on our results @xcite . [ cols="^,^ " , ] figure [ fig1 ] shows some results demonstrating the power of this combination of lhc analyses to constrain the set of pmssm neutralino lsp models . in particular
this figure shows the fraction of models with a given sparticle and lsp mass which are excluded by the combined lhc searches . since the masses of the squarks and gluinos ,
the lightest stops and sbottoms and the lsp itself are currently of particular interest , we have concentrated on those quantities in this figure . in the upper left - hand panel
the coverage of the gluino - lsp mass plane by the lhc searches is displayed ; the solid white line represents the simplified model @xmath9 cl search limit from atlas @xcite which we see is roughly the same as the all - black region excluded in the pmssm .
the upper right - hand panel shows the corresponding coverage in the gluino - lightest squark mass plane with the analogous simplified model line from atlas @xcite ( which assumes a zero mass lsp and that the 8 squarks of the first two generations are degenerate , neither of which are necessarily true in the pmssm ) .
while most models with rather light squarks and/or gluinos are observed to be excluded by the lhc searches , it is clear that models with both squarks and gluinos below @xmath10 gev still remain viable .
the lower two panels show the lightest stop / sbottom vs. lsp mass plane with the corresponding atlas simplified model search limits @xcite . here
we see several things , the most important being that the amount of coverage of these two mass planes in the pmssm by the lhc searches is significantly larger than what we would have expected from either simplified model analysis .
we also find that the third generation searches are ineffective for models that have a relatively small splitting between the stop and lsp ( @xmath11 130 gev ) , so the exclusion of these models relies almost entirely on the standard jets + met searches . in the uncompressed regions , on the other hand , the third generation searches significantly outperform jets + met , as expected .
in particular , the @xmath12+met analysis is found to be responsible for nearly all of the exclusion power in this region .
clearly , the third - generation searches cover important gaps in the standard jets + met coverage . in general , the third - generation searches are most sensitive to models in which stop or sbottom decays produce hard @xmath13-jets , and have a more difficult time observing models with stops or sbottoms that produce tops or decay to multiple intermediate gauginos ( resulting in softer @xmath13-jets and leptons ) .
lastly we again stress that the direct sbottom search is found to be responsible for the majority of the heavy flavor search exclusion power for models with relatively light stops .
cl limit results obtained by atlas in the simplified model limit as discussed in the text.,title="fig:",width=336 ] cl limit results obtained by atlas in the simplified model limit as discussed in the text.,title="fig:",width=336 ] cl limit results obtained by atlas in the simplified model limit as discussed in the text.,title="fig:",width=336 ] cl limit results obtained by atlas in the simplified model limit as discussed in the text.,title="fig:",width=336 ] in addition to the 7 and 8 tev lhc searches , future data taking and enhanced analyses at @xmath1 14 tev will greatly extend the expected coverage of the pmssm parameter space for both lsp types . in @xcite
, we consider the impact of one of the most powerful of these searches to be performed by atlas - the zero - lepton jets + met final state @xcite - following the same procedure that was used for the 7/8 tev analyses .
note that in extrapolating from 300 fb@xmath14 to 3 ab@xmath14 luminosity scaling has been employed to obtain the expected limit . as a result of limitations on cpu time
, we only generated 14 tev results for the @xmath15k neutralino lsp models that survive the 7 and 8 tev lhc analyses and also predict a higgs mass of @xmath16 gev .
we note that since the results of the 7 and 8 tev analyses are essentially independent of the higgs mass , it is quite likely that our results for this narrow higgs mass range would in fact be applicable , at least to a very good approximation , to the entire neutralino lsp model set .
our results for the atlas 14 tev zero - lepton jets+met search are summarized in fig .
[ fig2 ] for both 0.3 and 3 ab@xmath14 of integrated luminosity in the gluino - lightest squark mass plane . here
we see that for typical models , the 14 tev lhc will be able to exclude squarks up to @xmath1 1.6 tev and gluinos up to @xmath1 2.7 tev , with only 7.9 ( 2.5 ) percent of models surviving after 300 ( 3000 ) fb@xmath14 of data .
a few models are seen to survive with much lighter squark and/or gluino masses ; in most cases these models survive by producing high-@xmath17 leptons and therefore falling outside of the 0-lepton search region , so adding other searches with leptons in the signal region will undoubtedly exclude many of the surviving models with light colored sparticles . to fully understand the capabilities of the 14 tev lhc will of course require a far more realistic study than is presently available since the lhc collaborations themselves are still unsure of how their detectors will perform at 14 tev with very high pileup conditions , and we can currently only simulate one of the 14 tev search channels ( although this search is the most powerful at 7/8 tev and is expected to dominate the 14 tev exclusions as well ) .
( left ) and 3000 fb@xmath14 ( right ) , in the lightest squark - gluino mass plane.,title="fig:",width=336 ] ( left ) and 3000 fb@xmath14 ( right ) , in the lightest squark - gluino mass plane.,title="fig:",width=336 ]
the direct detection of dm results from either the spin - independent ( si ) or spin - dependent ( sd ) scattering of the lsp by a target nucleus . while @xmath18 @xmath19channel exchange only contributes to the sd(si ) process at tree level , @xmath20channel squark exchange can contribute to both processes .
however , as the lower bounds on the first and second generation squark masses from the lhc become stronger the importance of these squark exchange contributions clearly become subdominant .
, @xmath21 and @xmath22 squarks are quite different , both sd and si interactions may have substantial isospin - dependent contributions so that @xmath23 and @xmath24 can be significantly different . ]
the @xmath25-exchange graph is sensitive to the higgsino content of the lsp whereas the higgs exchange graph probes the product of the lsp s gaugino and higgsino content .
similarly , ( valence ) squark exchange is sensitive to the lsp s wino and bino content .
figure [ fig3 ] displays the predicted experimentally observable si and sd cross sections for our pmssm model set together with several present and anticipated future experimental constraints .
note that the actual cross sections are appropriately scaled by the factor @xmath26 to account for the fact that most of our pmssm models lead to a thermal relic density somewhat below that which saturates the wmap value as discussed in the introduction .
there are several things to note in this figure : ( @xmath3 ) future si searches will cut rather deeply into the model set ; a lack of signal at xenon1 t ( lz ) would exclude 23% ( 50% ) of these models .
_ however _ , this implies that about half of our models are not accessible to si experiments due to their rather small scaled cross sections , @xmath27 .
models with these values of @xmath27 tend to produce these small values due to both the suppression arising from their low thermal relic density as well as fact that the lsps tend to be rather close to being pure weak eigenstates as discussed above .
note that if we only consider models which predict a relic density within 10% of the critical density , the situation improves significantly , with 60% ( 87% ) of models within the xenon1 t ( lz ) exclusion limit .
( @xmath4 ) sd measurements are still rather far from the pmssm model predictions and we do not expect future sd measurements to have a significant impact on the parameter space ; sd experiments such as coupp500 will only be able to exclude @xmath28 of the models in this set if no signal is seen . while direct detection experiments have significant power to cover much of the pmssm , clearly such experiments will need to be ` supplemented ' if we want to discover or exclude the full range of neutralino lsps in the pmssm model space . in the next two sections , we describe constraints on the pmssm from indirect detection and neutrino telescope experiments .
indirect detection plays a critical role in searches for dm and , in the case of null results , can lead to very strong constraints on the pmssm parameter space . as will be seen below , both _
fermi _ and cta can contribute in different regions of the pmssm parameter space in the future .
cta , in particular , will be seen to be extremely powerful in the search for heavy lsps which are mostly higgsino- or wino - like and that predict thermal relic densities within an order of magnitude of the wmap / planck value . _
fermi _ , on the other hand , will be seen to be mostly sensitive to well - tempered neutralinos that are relatively light . the most promising dm targets for both _ fermi
_ lat and cta are those with both a high dm density and low astrophysical gamma - ray foregrounds .
these criteria have motivated a number of galactic and extragalactic targets including the galactic center ( gc ) , dwarf spheroidal galaxies ( dsphs ) , and galaxy clusters .
the expected gamma - ray signal for dm annihilations is proportional to the integral of the square of the dm density along the line of sight to the source ( @xmath29 ) .
the determination of @xmath29 is most reliable for dm - dominated objects such as dsphs and galaxy clusters in which the dm distribution can be robustly measured . in the milky way halo the uncertainty on the dm distribution rapidly increases as one approaches the inner galaxy where baryons dominate the gravitational potential .
kinematic data at large scales constrain the density of dm at the solar radius to 0.20.4 gev @xmath30 @xcite . under well - motivated models for the dm distribution in the galactic halo ,
the gc is expected to be the most intense dm source in the sky .
cdm simulations predict that the galactic dm halo should have a density profile with an inner cusp , @xmath31 with @xmath32 . for an extrapolation of the galactic dm density profile with @xmath33 , the expected dm signal from the gc region
is approximately two orders of magnitude greater than dsphs or galaxy clusters .
however the gc is also the region of the sky with the highest density of gamma - ray sources and brightest diffuse gamma - ray emission produced from the interaction of cosmic rays with the interstellar medium .
these foregrounds significantly limit the sensitivity of the _ fermi_-lat in the inner galaxy and complicate the interpretation of any observed signals . for our study of _
we focus on the sensitivity to the dm signal from dsphs which are predominantly at high galactic latitudes where the astrophysical foregrounds are much weaker . above 50 gev the diffuse emission from the galaxy is much less intense relative to other backgrounds than in the _ fermi_-lat sensitivity band .
the inner galaxy is thus the preferred target for cta under the assumption that the galactic dm halo possesses an inner cusp with @xmath32 .
here we follow the procedure developed in ackermann et al .
( henceforth a11 ) @xcite and extended in cotta et al .
@xcite to constrain the annihilation cross section , @xmath34 , for each pmssm model using _
fermi_-lat observations of ten dwarf spheroidal galaxies ( dsphs ) .
our two - year @xmath35-ray event sample is identical to that described in a11 , accepting photons in the energy range from @xmath36 within 10@xmath37of each dsph . in accord with a11 , we use the lat sciencetools version v9r20p0 and the p6_v3_diffuse instrument response functions .
j - factors and associated statistical uncertainties for the dsphs are taken from table 1 of a11 , where they were calculated using line - of - sight stellar velocities and the jeans equation @xcite .
similar to cotta et al .
, we use darksusy 5.0.5 @xcite to model the @xmath35-ray spectrum from the annihilation of each pmssm lsp .
darksusy calculates the total @xmath35-ray yield from annihilation , as well as the rates into each of 27 final state channels .
we calculate a joint likelihood to constrain the annihilation cross section of each pmssm model given the lat observations coincident with the ten dsphs .
following a11 , we incorporated statistical uncertainties in the j - factors of the dsphs as nuisance parameters in our likelihood formulation ( see equation 1 and the associated discussion in a11 ) .
this likelihood formulation includes both the flux normalizations of background @xmath35-ray sources ( diffuse and point - like ) and the associated dsph j - factors and statistical uncertainties .
no significant @xmath35-ray signal is detected from any of the dsphs when analyzed individually or jointly for any of the pmssm models .
for each of the pmssm models , we calculate the maximum annihilation cross section , @xmath38 , consistent with the null detection in the lat data .
we incorporate nuisance parameters to obtain a 95% one - sided confidence interval on the value of @xmath34using the profile likelihood method @xcite .
this one - sided 95% confidence limit on @xmath34serves as our value of @xmath38 , which is compared to the true value of the annihilation cross section for each pmssm model .
we define the `` boost '' necessary to constrain a model as the ratio @xmath39 . while the lat data do not presently constrain any of the pmssm models , it is useful to estimate the improvements expected over a 10 year mission lifetime . in the low - energy ,
background dominated regime , the lat point source sensitivity increases as roughly the square - root of the integration time .
however , in the high - energy , limited background regime ( where many pmssm models contribute ) , the lat sensitivity increases more linearly with integration time .
thus , 10 years of data could provide a factor of @xmath40 to 5 increase in sensitivity .
additionally , optical surveys such as pan - starrs @xcite , the dark energy survey @xcite , and lsst @xcite could provide a factor of 3 increase in the number of milky way dsphs corresponding to an increased constraining power of @xmath41 to 3 @xcite .
ongoing improvements in lat event reconstruction , a better understanding of background contamination , and an increased energy range are all expected to provide additional increases in the lat sensitivity .
thus , we find it plausible that the lat constraints could improve by a factor of 10 compared to current constraints .
in figure [ fig4 ] we display the boost required to constrain the various pmssm models at @xmath9 cl based on the _
fermi_-lat dwarf analysis employing only the first 2 years of data colored - coded by either the annihilation cross section or the lsp thermal relic density .
here we see that the lat analysis does not currently constrain any of our pmssm models .
however , as discussed above with more dwarfs and longer integration times we would expect an @xmath42-fold improvement in the sensitivity and thus all models with boost factors less than 10 would become accessible .
we will assume this @xmath42-fold improvement in sensitivity for the analysis that follows .
, for each model is plotted on the color scale .
( right ) here the corresponding relic density for each model is plotted on the color scale.,title="fig:",width=336 ] , for each model is plotted on the color scale .
( right ) here the corresponding relic density for each model is plotted on the color scale.,title="fig:",width=336 ] the cherenkov telescope array ( cta ) @xcite is a future ground - based gamma - ray observatory that will have sensitivity over the energy range from a few tens of gev to a few hundreds of tev . to achieve
the best sensitivity over this wide energy range cta will include three telescope types : large size telescope ( lst , 23 m diameter ) , medium size telescope ( mst , 10 - 12 m ) and small size telescope ( sst , 4 - 6 m )
. over this energy range the point - source sensitivity of cta will be at least one order of magnitude better than current generation imaging atmospheric cherenkov telescopes such as h.e.s.s . , magic , and veritas .
cta will also have an angular resolution at least 23 times better than current ground - based instruments , improving with energy from 0.1@xmath43 at 100 gev to better than 0.03@xmath43 at energies above 1 tev .
the optimal dm search region for cta will be limited by the cta fov of 8@xmath43 to the area within @xmath44 to @xmath45 of the gc . on these angular scales
the dm signal is predominantly determined by the dm distribution in the inner galaxy ( @xmath46 kpc ) .
we model the galactic dm distribution with an nfw profile with a scale radius of 20 kpc normalized to 0.4 gev @xmath30 at the solar radius .
this model is consistent with all current observational constraints on the galactic dm halo and represents a conservative expectation for the inner dm profile in the absence of baryons . because the annihilation signal is proportional to the square of the dm density , the projected limits for cta depend strongly on the assumptions that are made on the shape and normalization of the galactic dm halo profile .
the projected limits presented here could change by as much as a factor of 10 given these uncertainties .
the potential of cta for dm searches and testing other exotic physics has been studied in detail by @xcite using the projected performance of several alternative array configurations @xcite with 1837 msts and different combinations of ssts and lsts . for this study
we use the projected sensitivity of a candidate cta configuration with 61 msts on a regular grid with 120 m spacing @xcite .
this telescope layout is similar to the anticipated layout of the baseline mst array with 1825 telescopes with an additional us contribution of 36 msts .
this configuration has a gamma - ray angular resolution that can be parameterized as a function of energy as @xmath47 gev)@xmath48 and a total effective area above 100 gev of @xmath110@xmath49 m@xmath50 .
we define the gc signal region as an annulus centered on the gc that extends from [email protected]@xmath43 and calculate the sensitivity of cta for an integrated exposure of 500 hours that is uniform over the whole region . an energy - dependent model for the background in the signal region
is taken from a simulation of residual hadronic contamination .
the uncertainty in the background model is calculated for a control region with no signal contamination and a solid angle equal to five times the signal region ( 14.3 deg@xmath50 ) .
we compute the sensitivity of cta using a two - dimensional likelihood analysis that models the distribution of detected gamma - rays in energy and angular offset from the gc . following the same procedure as the _ fermi_-lat analysis , we compute the model boost factor as the ratio of the model cross section with the 95% c.l .
upper limit on the annihilation cross section of a dm model with the same spectral shape . for models with lsp masses below 1 tev ,
the sensitivity of cta is dominated by events at low - energy ( @xmath51 gev ) .
figure [ fig : cta_pmssm_sigmav ] shows the distribution of the cta boost factor versus lsp mass for all pmssm models and the subset of models that have an lsp relic density consistent with 100% of the dm relic density .
cta can exclude @xmath120% of the total model set and @xmath5250% of the models in the subset with the dm relic density .
here we see that @xmath53 of the models would be excluded by cta if no signal were to be observed .
cta boost factor vs. lsp mass for the full pmssm model set ( left ) and pmssm models satisfying current experimental constraints and a lsp relic density consistent with the wmap7 dm density .
the true cross section , @xmath34 , for each model is plotted on the color scale.,title="fig:",scaledwidth=49.0% ] cta boost factor vs. lsp mass for the full pmssm model set ( left ) and pmssm models satisfying current experimental constraints and a lsp relic density consistent with the wmap7 dm density .
the true cross section , @xmath34 , for each model is plotted on the color scale.,title="fig:",scaledwidth=49.0% ]
neutralino dark matter can be captured and accumulated in the sun .
neutralinos in this relatively overdense population would then also sink to the solar core and annihilate .
if the product of capture and annihilation cross sections is large enough this process leads to an equilibrium population of captured neutralinos whose annihilations are proportional to their _ elastic scattering cross - sections _
@xcite and that may be detectable by observing an excess of high - energy ( @xmath54 ) solar neutrinos in @xmath55-scale neutrino telescope experiments @xcite . here
we present predictions for an icecube / deepcore ( ic / dc ) search for neutralino dm in our model set .
our analysis closely follows that presented in @xcite . in the results presented here
we assume that each neutralino s relic density is given by the usual thermal calculation .
we use darksusy 5.0.6 @xcite to simulate the ( yearly - average ) signal @xmath56/@xmath57 neutrino flux spectra incident at the detector s position and convolve with preliminary @xmath56/@xmath57 effective areas for muon events contained in deepcore .
we consider a data set that includes @xmath58yr of data that is taken during austral winters ( the part of the year for which the sun is in the northern hemisphere ) over a total period of @xmath42yrs .
an irreducible background rate of @xmath42 events / yr is expected from cosmic ray interactions with nuclei in the sun . here
we will take ( as discussed at greater length in @xcite ) a detected flux of @xmath59 events / yr as a conservative criterion for exclusion .
the basic results of this analysis are presented in figure [ icdc ] . in figure
[ icdc ] , we show all pmssm models in our set using gray points , while highlighting wmap - saturating models with mostly bino , wino , higgsino or mixed ( @xmath6080% of each ) lsps in red , blue , green and magenta , respectively .
detectability is tightly correlated with the elastic scattering cross - sections ( @xmath61 and @xmath62 ) while having little correlation with the annihilation cross - section @xmath63 , as expected .
the biggest difference between these results and those of the previous analysis @xcite , which used a set of pmssm models that were chosen to have relatively light ( @xmath64 ) sparticles , is that a much smaller percentage of the current pmssm models are able to reach capture / annihilation equilibrium in the sun .
this is due to the fact that so many of these models are nearly pure wino or higgsino gauge eigenstates ( which have both low relic density and small capture cross - sections ) and that the lsps in this model set tend to be much heavier than those in the previous set . if one defines out - of - equilibrium models as those with solar annihilation rates less than 90% of their capture rates , we find that no such models can be excluded by ic / dc .
in contrast , relatively light lsps composed of a mixture of gaugino and higgsino eigenstates have large scattering and annihilation cross sections and are highly detectable by ic / dc .
we observe that all such wmap - saturating well - tempered neutralinos with masses @xmath65 should be excluded by the ic / dc search ( _ c.f .
_ , the magenta points in fig . [ icdc ] ) .
( upper - right ) and thermal elastic scattering cross - sections @xmath66 and @xmath67 ( lower panels ) . in all panels the gray points represent generic models in our full pmssm model set , while wmap - saturating models with mostly bino , wino , higgsino or mixed ( @xmath6080% of each ) lsps in are highlighted in red , blue , green and magenta , respectively .
the red line denotes a detected flux of @xmath68 events / yr , our conservative estimate for exclusion.,scaledwidth=100.0% ]
now that we have provided an overview of the various pieces of data that go into our analysis , we can put them together to see what they ( will ) tell us about the nature of the neutralino lsp as dm @xcite and , more generally , the pmssm itself .
we remind the reader that this is an ongoing analysis and that several future updates will be made to what we present here before completion .
since we only have 14 tev results for the @xmath1 30.7k neutralino lsp models that survive the 7 + 8 tev searches and have @xmath7 gev ( because of cpu limitations as described above ) , the main results presented below will use only the 7 + 8 tev lhc searches listed in table [ searchlist ] .
we will also present some indicative results showing the sensitivity of the combined 7,8 , and 14 tev lhc analyses for the subset of neutralino lsp models with @xmath7 gev .
[ figxx ] shows the survival and exclusion rates resulting from the various searches and their combinations in the lsp mass - scaled si cross section plane . in the upper left panel
we compare these for the combined direct detection ( dd = xenon1 t + coupp500 ) and indirect detection ( i d = _ fermi
_ + cta ) dm searches . here
we see that 11% ( 15% ) of the models are excluded by i d but not dd ( excluded by dd but not i d ) while 8% are excluded by both searches . on the other hand , we also see that 66% of the models survive both sets of dm searches ; 44% of this subset of models , in turn , are excluded by the lhc searches .
note that the dd- and id - excluded regions are all relatively well separated in terms of mass and cross section although there is some overlap between the sets of models excluded by the different experiments .
in particular we see that the i d searches ( here almost entirely cta ) are covering the heavy lsp region even in cases where the si cross section is very low and likely beyond the reach of any potential dd experiment .
similarly , in the upper right panel we see that 22% ( 0% ) of the models are excluded uniquely by dd ( icecube ) only while 1% can be simultaneously excluded by both sets of searches and 77% would be missed by either search set . in the lower left panel ,
i d and lhc searches are compared and we see that 17% ( 35% ) of the models would be excluded only by the i d ( lhc ) searches . however , 2% ( 47% ) of the models are seen to be excluded by ( or would survive ) both searches .
the strong complementarity between the lhc , cta and xenon1 t experiments is evident here as cta probes the high lsp mass region very well where winos and higgsinos dominate , xenon1 t chops off the top of the distribution where the well - tempered neutralino lsp states dominate , and the lhc covers the relatively light lsp region ( independent of lsp type ) rather well .
of course the strength of the lhc coverage will significantly improve when 14 tev searches are included , as we will see below . in the lower right panel
the relative contributions arising from the lhc and cta searches to the model survival / exclusion are shown . here
the intensity of a given bin indicates the fraction of models excluded there by the combination of both cta and the lhc while the color itself indicates the weighting from cta ( blue ) and lhc ( red ) .
it is again quite clear that cta completely dominates for large lsp masses ( which correspond to the mostly wino and higgsino lsps ) and also competes with the lhc throughout the band along the top of the distribution where the lsp thermal relic density is approximately saturating wmap / planck .
the lhc covers the rest of the region but is not yet as effective as cta ( excluding a smaller fraction of models in the relevant bins ) . in order to study the complementarity arising from the various searches in more detail
, we found that it is very useful to project their individual capabilities onto several sets of fixed parameter planes that are directly related to one of the search categories .
we begin these projections with fig .
[ figyy1 ] which compares the various search capabilities in the familiar lsp mass - scaled si cross section plane , most relevant for the si dd searches , as defined by the fraction of models excluded in the various bins .
here we see the contributions to the search coverage coming uniquely from cta ( top left ) , the lhc ( top right ) , icecube ( bottom left ) and _ fermi _ ( bottom right ) and also show the limit that could arise from xenon1 t for comparison . here
we see that both icecube and _ fermi _ probe models with low lsp masses and large si cross sections , where the lsp tends to be a bino - higgsino admixture ; this region is also accessible to the dd experiments such as xenon1 t . on the other hand
, cta has access to the heavy lsp region , where there are a large fraction of relatively pure winos and higgsinos , while the lhc coverage is concentrated ( for now ) on the relatively low mass lsp region .
interestingly , the lhc searches are not independent of the si cross - section ; a region of enhanced exclusion fraction is seen for si cross sections near @xmath69 pb . models with cross sections in this region mostly have wino - like lsps with light squarks ; wino - like lsps with heavier squarks have a lower si cross - section , while higgsinos and mixed states tend to have a higher si cross - section whether or not light squarks are present .
annihilation into @xmath70 final states only.,title="fig:",width=336 ] annihilation into @xmath70 final states only.,title="fig:",width=336 ] annihilation into @xmath70 final states only.,title="fig:",width=336 ] annihilation into @xmath70 final states only.,title="fig:",width=336 ] taking these same results we can now project them in to the plane which is most relevant for dm i d searches . fig .
[ figyy2 ] compares the various search capabilities as above but now in the lsp mass - lsp pair annihilation cross section plane , for which both _ fermi _ and cta are directly applicable . here
we see the fraction of models that would be excluded by searches at cta ( top left ) , the lhc ( top right ) , icecube ( bottom left ) and xenon1 t ( bottom right ) . the potential bounds from _ fermi _ and cta are also shown in all cases as curves penetrating through the upper left and right parts of the figures , respectively . here
the dashed ( solid ) curves correspond to the limit obtained when the lsp pair annihilation final state is exclusively @xmath70 ; we emphasize that a generic lsp in a pmssm model may annihilate to many different final states beyond these two cases .
note again that cta is primarily sensitive to models with lsp mass above 100 gev and with @xmath34relatively close to the thermal relic value .
as in the previous figure , the lhc is mainly effective in the lower lsp mass region .
in addition , the lhc searches are seen to be particularly efficient along a thin line starting at annihilation cross sections of @xmath71 for 100 gev lsps and increasing with lsp mass ; this line corresponds to models with a pure wino lsp , which are subject to strong constraints from searches for heavy stable charged particles at the lhc .
( note that this line is below the projected reach of the i d searches for lsp masses accessible to the current lhc searches . )
icecube is seen to nicely complement cta in the region with large cross sections but low lsp masses .
the projection of the xenon1 t coverage onto this plane is interesting with most of the exclusion appearing at lower lsp masses where the lsp pair annihilation cross section is also large .
however , there is also an extended , somewhat diffuse region of substantial coverage throughout the entire right half of the parameter space as well as on the @xmath72 funnel ` island ' at small lsp masses . as a final example of these projections , we now consider the thermal relic density - lsp mass plane ;
the relic density plays a ` linear ' role in the dd searches but a ` quadratic ' one in the i d case ( and none at all as far as the lhc is concerned ) . in fig .
[ figyy3 ] we display the coverage comparisons for the various experiments : cta , lhc , icecube , xenon1 t and _ fermi _ , respectively .
again we see that multiple searches can make overlapping contributions in various parts of this parameter space plane .
these plots show several things : first , cta , as expected , does an excellent job at accessing most of the models with lsp masses above @xmath73 gev that saturate the relic density .
however , for larger masses the cta coverage extends down to relic densities as much as a factor of @xmath42 or so below the wmap / planck value .
_ fermi _ is seen to cover only the low lsp mass region with relic densities not far from the thermal value while icecube can go to much lower relic densities provided the lsp mass is below @xmath74 gev or so .
xenon1 t displays coverage throughout this plane but does best for lsp masses below @xmath75 gev , providing coverage even for models with very low relic densities .
of course , even for masses up to 1 - 2 tev , xenon1 t still provides quite decent model coverage in this parameter plane . as noted already , most of the impact of the lhc is at present seen to be at lower lsp masses below @xmath74 gev .
the lhc coverage is relatively uniform as far as the value of the relic density is concerned except in the case of very light lsps where the coverage is very strong . of course
, we again remind the reader that extending the lhc energy to 14 tev will substantially improve its effectiveness for excluding heavy lsps . .
, width=672 ] finally , fig .
[ figzz ] shows the impact of combining all of the different searches in this same @xmath76-lsp mass plane which should be compared with that for the original model set as generated that is shown in fig .
[ fig00 ] . here
we see that ( @xmath3 ) the models that were in the light @xmath77 and @xmath25-funnel regions have essentially evaporated ( and further measurements of the invisible width of higgs as well as the standard susy searches will likely further restrict these ) , ( @xmath4 ) the well - tempered neutralinos are now seen to be completely gone , mostly due to xenon1 t and icecube ; ( @xmath5 ) the possibility of almost pure higgsino or wino lsps even approximately saturating the relic density has vanished thanks to cta and ( @xmath78 the mixed wino - higgsino models , due to a combination of measurements , have also completely disappeared .
( @xmath79 ) the only models remaining which _ do _ saturate the wmap / planck relic density are those with binos with resonant or co - annihilations .
( @xmath80 ) we find that @xmath81 of all the pmssm models have been excluded by at least one of the searches considered in this paper .
we now consider the effect of adding the 14 tev jets + met search with 300 fb@xmath14 of integrated luminosity to the full set of 7 + 8 tev searches , considering only the subset of models with @xmath82 gev .
we note that both the lhc and dark matter results are essentially independent of the higgs mass , so that the results for this subset should reproduce the results we would get for the full model set , albeit with lower statistics .
the left panel of figure [ 14tevcomp ] shows the reach of the combined lhc searches in the lsp mass - si cross - section plane for models with @xmath82 gev . comparing this figure to the upper right panel of figure [ figyy1 ]
, we see two key changes with the inclusion of the 14 tev search .
first , in both cases the effectiveness of the lhc searches is seen to fall off sharply above a particular lsp mass , since above this limit the spectrum is generically either too heavy or too compressed to observe . adding the 14 tev searches effectively doubles this cutoff , from @xmath1 700 gev to @xmath1 1400 gev , so that most lsps in our model set are now accessible given colored sparticle masses below @xmath1 2 - 3 tev .
second , we see that the lhc now excludes a very high fraction ( but not all ) of the models with lsps lighter than the aforementioned cutoff .
the large fraction of models which are excluded with 14 tev data is unsurprising , since our chosen upper limits for the sparticle masses were designed to ensure that most models would be ( at least kinematically ) accessible at the 14 tev lhc . for models with the correct higgs mass , shown in the lsp mass - scaled si cross section plane ( left panel ) and a comparison between the fractions excluded by the lhc and cta in this plane ( right panel ) .
the potential si bound from xenon1 t is also shown in all cases.,title="fig:",width=336 ] for models with the correct higgs mass , shown in the lsp mass - scaled si cross section plane ( left panel ) and a comparison between the fractions excluded by the lhc and cta in this plane ( right panel ) .
the potential si bound from xenon1 t is also shown in all cases.,title="fig:",width=336 ] the right panel of figure [ 14tevcomp ] shows a comparison between the reaches of the lhc and cta in this plane , analogous to the lower right panel of figure [ figxx ] .
we see that the lhc and cta sensitivities now exhibit a sizable region of overlap .
however , the blue region on the far right edge of the plot shows that cta will still be sensitive to lsp masses beyond the reach of the lhc .
we also note that lsps heavy enough to be seen by cta are generally too heavy to be seen in direct ( e.g. monojet ) searches , meaning that the lhc is excluding these models by observing other ( mostly colored ) sparticles .
it is therefore important to note that cta has the potential to exclude winos and higgsinos with nearly thermal relic densities regardless of the characteristics of the rest of the sparticle spectrum .
of course , the improved reach of the lhc means that there is also an increasing overlap between the lhc and xenon 1 t , however we see that ( as before ) the lhc searches are mostly independent of the si cross - section ( the white areas around the edges of the plot generally result from low statistics in those regions ) . of course
, from a discovery perspective , the increased overlap means that there is more potential for a signal to be observed by two , or even all three , experiments , which would greatly aid in characterizing the lsp and model properties .
finally , we note that not only has the total fraction of models excluded by the combined experiments increased dramatically ( from 63% to 97.4% ) as a result of the improved lhc reach , but the fraction of models not seen by the lhc which are excluded by direct or indirect detection has increased ( from 41% to 49% ) , since more of the undetected models have heavy lsps and are therefore likely to be excluded by cta .
the results presented here are preliminary but directly lead us to a number of interesting conclusions which are already quite clear and which we expect to strengthen in the future : * even if the lsp _ does not _ make up all of the dm , it can still be observed in both direct and indirect detection experiments as well as in neutrino experiments such as icecube .
of course , searches at the lhc are not influenced by the lsp relic density .
* the set of models remaining after all the searches are performed that saturate the thermal relic density consist only of those with ( co)annihilating bino lsps . * si direct detection , cta , and the lhc do most of the heavy lifting in terms of complementary searches covering the pmssm parameter space .
* multiple / overlapping searches allow for extensive parameter space coverage which will be of particular importance if a dm signal is observed . *
most of the experiments are seen to provide complementary probes of the pmssm parameter space .
* the strength of the lhc component in these searches increases significantly with the inclusion of the 14 tev lhc jets + met search .
however , the jets + met search is dependent on the rest of the model spectrum and therefore does not provide complete coverage of any given lsp scenario , in contrast to dm searches which rely more directly on the lsp properties . in summary
, the pmssm provides an excellent tool for studying complementarity between different approaches to the search for dark matter .
the authors would like to thank j. feng , k. matchev , and t. tait for useful discussions .
this work was supported by the department of energy , contract de - ac02 - 76sf00515 , the department of energy office of science graduate fellowship program ( doe scgf ) , made possible in part by the american recovery and reinvestment act of 2009 , administered by orise - orau under contract no .
de - ac05 - 06or23100 , and the national science foundation under grant phy-0970173 . for an overview of mssm physics and phenomenology as well as msugra ,
see m. drees , r. godbole , p. roy , hackensack , usa : world scientific ( 2004 ) 555 p ; h. baer , x. tata , cambridge , uk : univ . pr .
( 2006 ) 537 p ; s. p. martin , in * kane , g.l .
( ed . ) : perspectives on supersymmetry ii * 1 - 153 [ hep - ph/9709356 ] .
t. cohen and j. g. wacker , arxiv:1305.2914 [ hep - ph ] .
see , for example , a. armbruster , atlas collaboration , talk given at _
recontres de moriond : qcd and high energy interaction _ , lathuile , march 9 - 16 , 2013 .
m. w. cahill - rowley , j. l. hewett , s. hoeche , a. ismail and t. g. rizzo , eur .
j. c * 72 * , 2156 ( 2012 ) [ arxiv:1206.4321 [ hep - ph ] ]
. m. w. cahill - rowley , j. l. hewett , a. ismail and t. g. rizzo , phys .
d * 86 * , 075015 ( 2012 ) [ arxiv:1206.5800 [ hep - ph ] ] , arxiv:1211.1981 [ hep - ph ] , arxiv:1211.7106 [ hep - ph ] and in preparation m. cahill - rowley , j. l. hewett , a. ismail and t. g. rizzo , arxiv:1307.8444 [ hep - ph ] .
e. komatsu _ et al .
_ [ wmap collaboration ] , astrophys . j. suppl . * 192 * , 18 ( 2011 ) [ arxiv:1001.4538 [ astro-ph.co ] ] .
b. c. allanach , comput .
commun . * 143 * , 305 ( 2002 ) [ hep - ph/0104145 ] .
a. djouadi , j. l. kneur and g. moultaka , comput .
commun .
* 176 * , 426 ( 2007 ) [ arxiv : hep - ph/0211331 ] .
a. djouadi , m. m. muhlleitner and m. spira , acta phys .
b * 38 * , 635 ( 2007 ) [ hep - ph/0609292 ] . c. f. berger , j. s. gainer , j. l. hewett and t. g. rizzo , jhep * 0902 * ( 2009 ) 023 [ arxiv:0812.0980 [ hep - ph ] ] ; j. a. conley , j. s. gainer , j. l. hewett , m. p. le and t. g. rizzo , eur .
j. c * 71 * ( 2011 ) 1697 [ arxiv:1009.2539 [ hep - ph ] ] and [ arxiv:1103.1697 [ hep - ph ] ] .
s. chatrchyan _ et al . _
[ cms collaboration ] , phys .
b * 713 * , 68 ( 2012 ) [ arxiv:1202.4083 [ hep - ex ] ] .
r. aaij _ et al . _
[ lhcb collaboration ] , arxiv:1307.5024 [ hep - ex ] ; s. chatrchyan _ et al . _
[ cms collaboration ] , arxiv:1307.5025 [ hep - ex ] .
the analysis as presented here , however , employs the earlier results as given in r. aaij _ et al .
_ [ lhcb collaboration ] , phys . rev .
* 110 * , 021801 ( 2013 ) [ arxiv:1211.2674 [ unknown ] ] .
the numerical impact of these later results on our analysis is very minor .
t. sjostrand , s. mrenna and p. z. skands , jhep * 0605 * , 026 ( 2006 ) [ hep - ph/0603175 ] .
j. conway , pgs4 , pretty good detector simulation , http://www.physics.ucdavis.edu/ conway / research / software / pgs / pgs.html .
k. bernlohr , a. barnacka , y. becherini , o. b. bigas , e. carmona , p. colin , g. decerprit and f. di pierro _ et al .
_ , astropart .
* 43 * , 171 ( 2013 ) [ arxiv:1210.3503 [ astro-ph.im ] ] .
jogler , t. , wood , m. d. , dumm , j. , & cta consortium 2012 , american institute of physics conference series , 1505 , 765 w. h. press and d. n. spergel , astrophys .
j. * 296 * , 679 ( 1985 ) .
a. gould , astrophys . j. * 321 * , 571 ( 1987 ) .
h. silverwood , p. scott , m. danninger , c. savage , j. edsj , j. adams , a. m. brown and k. hultqvist , jcap * 1303 * , 027 ( 2013 ) [ arxiv:1210.0844 [ hep - ph ] ] .
r. c. cotta , k. t. k. howe , j. l. hewett and t. g. rizzo , phys .
d * 85 * , 035017 ( 2012 ) [ arxiv:1105.1199 [ hep - ph ] ] . c. boehm , p. s. b. dev , a. mazumdar and e. pukartas , jhep * 1306 * , 113 ( 2013 ) [ arxiv:1303.5386 [ hep - ph ] ] . | the search for and identification of neutralino dark matter in supersymmetry requires a multi - pronged approach with important roles played by collider , direct and indirect dark matter detection experiments . in this report
, we summarize the sensitivity of such searches at the 7 , 8 ( and eventually 14 ) tev lhc , combined with those by _
fermi _ , cta ,
icecube / deepcore , coupp and xenon1 t , to such particles within the context of the 19-parameter p(henomenological)mssm .
this report provides an outline of the current status of our results and our expectations for future analyses . |
pulse shape discrimination ( psd ) is a fundamental technique for background rejection in scintillator detectors .
it relies on the different time profiles of the emitted scintillator light to solicitations from particles with different energy losses . for this reason , it is particularly effective for discriminating heavy particles ( ions , alphas or protons ) from light particles ( electrons , positrons ) .
even so , electrons and positrons generate similar light pulses and therefore psd fails in discriminating them .
psd discrimination between electrons and positrons can be recovered exploiting the deformation of the positron induced light pulse shape caused by the formation of positronium , a meta - stable positron - electron bound state .
depending on the system total spin , the positronium can be formed in the singlet state , called para - positronium ( p - ps , b.r .
25@xmath0 ) , or in the triplet state , called ortho - positronium ( o - ps , b.r .
75@xmath0 ) . due to the charge - conjugation invariance
, p - ps decays into two 511 kev gammas , with a mean life in vacuum of 125 ps , where o - ps decays into three gammas of total energy equal to 2@xmath1 , with a mean life in vacuum of 142 ns .
nevertheless , in matter the o - ps mean life is strongly reduced by the interactions with the surrounding medium .
chemical reactions ( oxidation , compound compositions ) , magnetic effects ( spin flip ) or pick - off ( positron annihilation with an anti - parallel spin electron of the medium ) cause the o - ps to decay into two gammas with a mean life of a few ns .
the decay into three gammas is usually reduced to a negligible fraction .
an o - ps mean life of a few ns is not enough for separating the positron energy deposition from the annihilation gammas in scintillator detectors , but it can still induce a positron pulse shape deformation strong enough to enhance electron - positron discrimination . this enhancement could be exploited in reactor neutrino experiments @xcite , in which anti - neutrino detection is performed via inverse beta decay @xmath2 the positron energy deposition and its following annihilation provide the prompt event and the neutron absorption the delayed event .
an o - ps enhanced psd could allow to suppress the accidental background and to reject the residual correlated background , given by the cosmogenic @xmath3n emitters @xmath4li and @xmath5he .
the same technique can be also exploited for background rejection in underground low background experiments , like borexino @xcite and sno+ @xcite , in which several @xmath6 emitters ( @xmath5b , @xmath4be , @xmath7c ) produced by cosmic muon spallation , constitute an unavoidable source of background .
it has already been successfully applied in borexino for the detection of solar _ pep _ neutrinos , where the rejection of cosmogenic @xmath8c is essential for the signal extraction .
the characterization of o - ps properties , i.e. the measurement of its formation probability and lifetime , has been performed for the most commonly used liquid scintillators , namely pc , lab , pxe and oil , and for the mixture pc + ppo @xcite .
furthermore , the effect of scintillator doping on the o - ps properties has been studied for different dopants : gd , commonly used in reactor neutrino experiments to enhance the neutrino signal identification , and nd , originally considered as neutrinoless double beta decay signal source in the sno+ experiment .
forthcoming measures will characterize also li - loaded and tl - loaded liquid scintillators , the latter being chosen as new dopant for the sno+ scintillator .
the o - ps properties ( lifetime and formation probability ) in scintillator are measured by means of a 0.8 mbq @xmath9na source in a fast coincidence system .
the radio - isotope @xmath9na decays @xmath6 to @xmath9ne with a b.r . of 89.2@xmath0 ( q - value = 2.842 mev ) . in 99.944@xmath0 of cases ,
the decay is to an excited state of @xmath9ne , which de - excites emitting a 1.274 mev gamma .
the o - ps properties are extrapolated from the measurement of the time distance between the de - excitation gamma ( prompt signal ) and the positron annihilation gammas ( dealyed signal ) .
a droplet of inert solvent containing the source is deposited between two identical supports , 1 cm in radius , each composed by two 7.5 @xmath10 m thick kapton foils .
kapton is an aromatic polyimide material where positrons are known not to form positronium @xcite .
the source sandwich is placed into a vial containing the scintillator sample .
the vial is in turn positioned between two plastic scintillator detectors , each coupled to a pmt .
the threshold on the first detector is set to 900 kev to isolate the @xmath9ne de - excitation gamma , while the second detector selects gammas between 350 kev and 500 kev , in order to accept the annihilation gammas but reject the de - excitation gamma backscattering .
a constant fraction discriminator generates a fast timing signal at each trigger from the two detectors . a time to amplitude converter generates a signal linearly increasing with time , starting at the prompt signal and stopping at the delayed signal .
the output is digitized by a 4096 channels adc .
figure [ fig : fit ] shows an example of resulting time spectrum .
the system is calibrated using a @xmath11co source , and each adc channel is found to corresponds to 10.6 ps .
the resolution is measured with the fast coincidence ( 182 ps ) gammas from @xmath12bi and found to be 0.28 ns at fwhm .
positronium has been characterized in common liquid scintillators ( pc , pxe , lab , oil , pc + ppo ) at room temperature .
each sample has been flushed with nitrogen and then measured with a @xmath13 event statistics , repeatedly from 3 to 5 times , in order to account for possible variations in the experimental conditions ( temperature , oxygen concentration ) .
the obtained time distributions are fitted using the roofit toolkit @xcite , embedded in the root package and based on minuit .
the fit function is @xmath14 where @xmath15 is the step function and a@xmath16 and @xmath17 are the amplitude and mean life of the _ i_-th component : the first accounts for the effective contribution of direct annihilation and p - ps , while the second accounts for the o - ps .
the constant term accounts for accidentals .
the function @xmath18 is convoluted with the detector resolution function , the sum of two gaussians with the same mean value , but different sigmas ( @xmath19 = 110 ps and @xmath20 = 160 ps ) and relative weights ( g@xmath21 = 0.8 and g@xmath22 = 0.2 ) . the obtained normalized @xmath23 are in the range between 0.85 and 0.94 and the statistical errors are at the per mil level . the o - ps mean life
is given directly by the fit parameter @xmath24 , while the formation probability @xmath25 is computed as the fraction of o - ps relative to the annihilations in the scintillator @xmath26 the fraction of direct annihilations in kapton , a@xmath27 , has been extrapolated from dedicated measurements with different number of kapton layers and estimated to be ( 20.6 @xmath28 0.2)@xmath0 .
the systematic uncertainty is evaluated from the dispersion of the measured mean lives from the weighted average of the corresponding scintillator sample .
the distribution of all the dispersions is fitted to a gaussian , whose standard deviation is taken as the systematic error .
the same procedure is followed for the formation probability .
the resulting systematic errors are 0.03 ns and 0.5@xmath0 .
table [ tab : a ] reports the results for the gd - loaded and nd - loaded liquid scintillators at the different concentrations . both of the dopants show the same effect on the o - ps properties : the mean life remains stable under dopant concentration variations , while the formation probability decreases from @xmath2950@xmath0 at concentrations of the order of few tenths per mil to @xmath2940@xmath0 at concentrations around 0.5@xmath0 .
y. abe _ et al . _
[ double chooz collaboration ] , phys .
d * 86 * , 052008 ( 2012 ) f. p. an _ et al . _ [ daya - bay collaboration ] , phys .
lett . * 108 * , 171803 ( 2012 ) j. k. ahn _
[ reno collaboration ] , phys .
lett . * 108 * , 191802 ( 2012 ) g. alimonti et al .
instrum .
a * 600 * , 568 ( 2009 ) | pulse shape discrimination in liquid scintillator detectors is a well - established technique for the discrimination of heavy particles from light particles .
nonetheless , it is not efficient in the separation of electrons and positrons , as they give rise to indistinguishable scintillator responses .
this inefficiency can be overtaken through the exploitation of the formation of ortho - positronium ( o - ps ) , which alters the time profile of light pulses induced by positrons .
we characterized the o - ps properties in the most commonly used liquid scintillators , i.e. pc , pxe , lab , oil and pc + ppo .
in addition , we studied the effects of scintillator doping on the o - ps properties for dopants currently used in neutrino experiments , gd and nd .
further measurements for li - loaded and tl - loaded liquid scintillators are foreseen .
we found that the o - ps properties are suitable for enhancing the electron - positron discrimination .
address = laboratoire astroparticule et cosmologie , paris address = department of aerospace science and technology , politecnico di milano address = laboratoire astroparticule et cosmologie , paris address = brookhaven national laboratory address = iphc , universit de strasbourg address = iphc , universit de strasbourg address = laboratoire astroparticule et cosmologie , paris address = brookhaven national laboratory |
most of the cr antiprotons observed near the earth are secondaries produced in collisions of energetic cr particles with interstellar gas @xcite . due to the kinematics of the process
, the spectrum of antiprotons has a unique shape distinguishing it from other cosmic - ray species : it is expected to peak at about 2 gev decreasing sharply towards lower energies . in addition to secondary antiprotons there are possible sources of primary antiprotons such as candidate dark matter particles and evaporating black holes . despite numerous efforts and overall agreement on the secondary nature of the majority of cr antiprotons , published estimates of the expected flux significantly differ ( see e.g. fig .
3 in orito et al .
the major sources of uncertainties are : ( i ) incomplete knowledge of cross sections for antiproton production , annihilation , and scattering , ( ii ) parameters and models of particle propagation in the galaxy , and ( iii ) modulation in the heliosphere .
while the interstellar antiproton flux is affected only by uncertainties in the cross sections and propagation models , the final comparison with experiment can only be made after correcting for the solar modulation .
besides , the spectra of cr nucleons have been directly measured only inside the heliosphere while we need to know the spectrum outside in interstellar space to compute the antiproton production rate correctly . we have developed a numerical method and corresponding computer code galprop for the calculation of galactic cr propagation in 3d @xcite .
the code has been shown to reproduce simultaneously observational data of many kinds related to cr origin and propagation @xcite .
the code has been validated on direct measurements of nuclei , antiprotons , electrons , and positrons , and astronomical measurements of and synchrotron radiation .
these data provide many independent constraints on model parameters . here
we use the galprop code for accurate calculation of production and propagation of secondary antiprotons .
we explore the dependence of the antiproton flux on the nucleon injection spectrum and propagation parameters .
the antiproton production is calculated using the @xmath0 production cross section and dtunuc nuclear factors @xcite or the @xmath0 production cross section scaled appropriately with atomic numbers .
inelastic scattering producing `` tertiary '' antiprotons and `` secondary '' protons is taken into account . the calculated local interstellar spectrum ( lis )
is modulated using the steady - state drift model . for the calculation reported here
, we use a cylindrically symmetrical galactic geometry .
lcccccc & injection & & & + model & index@xmath1 , @xmath2 & @xmath3 , @xmath4 s@xmath5 & index , @xmath6 & & @xmath7 , km s@xmath5 & @xmath8 , km s@xmath5 kpc@xmath5 + stochastic reacceleration ( sr ) & 2.43 & @xmath9 & 0.33 & & 30 & + minimal reacceleration & convection ( mrc ) & 2.43 & @xmath10 & 0.33 & & 17 & 10 + plain diffusion ( pd ) & 2.16 & @xmath11 & 0.6 & & & + diffusion plus convection ( dc ) & @xmath12 & @xmath13 & @xmath14 & & & 10 +
the propagation parameters have been fixed using the b / c ratio .
nucleon injection spectra were chosen to reproduce the local cr measurements .
the source abundances of all isotopes @xmath15 are given in @xcite .
we thus use the same ( webber et al . )
cross - section parametrization as in that work in our calculations . in all cases
the halo size has been set to @xmath16 kpc , which is within the range @xmath17 kpc derived using the galrpop code and the combined measurements of radioactive isotope abundances , @xmath18be , @xmath19al , @xmath20cl , and @xmath21mn ( strong and moskalenko 2001 and references therein ) .
note that the exact value of @xmath22 is unimportant for antiproton calculations provided that the propagation parameters are tuned to match the b / c ratio .
the `` tertiary '' antiprotons ( inelastically scattered secondaries ) , significant at the lowest energies , are important in interstellar space , but make no difference when compared with measurements in the heliosphere . to investigate the range of interstellar spectra and propagation parameters we considered four basic models ( table [ table1 ] ) .
our results are plotted in figs . [
fig : protons]-[fig : pos ] . a model with the stochastic reacceleration ( sr ) reproduces the sharp peak in secondary to primary nuclei ratios in a physically understandable way without breaks in the diffusion coefficient and/or the injection spectrum @xcite .
however , this model produces a bump in proton and he spectra at @xmath23 gev / nucleon which is not observed .
this bump can be removed by choosing an injection spectrum that hardens at low energies @xcite .
there are however some problems with secondaries such as positrons and antiprotons that are more difficult to manage .
a similar bump appears in the positron spectrum at @xmath24 gev ( fig .
[ fig : pos ] ) , and the model underproduces antiprotons at 2 gev by more than 30% ( fig.[fig : pbars ] ) . taken together
they provide evidence against strong reacceleration in the ism .
another model combining reduced reacceleration and convection ( mrc ) also produces too few antiprotons . using a plain diffusion model ( pd ) we can get good agreement with b / c above few gev / nucleon , with nucleon spectra and positrons , but this model overproduces antiprotons at 2 gev by @xmath25% and contradicts the secondary / primary nuclei ratio ( b / c ) below 1 gev / nucleon .
mv ) . the line coding : solid lines
dc model ( thick line modulated spectrum ) , dashes dr , dots pd .
the upper curve is always lis spectrum , the lower is modulated .
data : imax @xcite , caprice @xcite , ams @xcite , bess @xcite . ]
s for 3 models . for each model ,
the upper curve is the lis , the lower curve modulated ( @xmath26 mv ) .
line coding as in fig .
[ fig : protons ] .
the two lowest curves show separately the lis and modulated spectrum of the `` tertiary '' component in dc model .
data : bess @xcite , mass91 @xcite , caprice98 @xcite . ]
a model including diffusion and convection ( dc ) is our best fitting model .
it reproduces all the particle data `` on average '' , although it has still some problem with the reproduction of the sharp peak in b / c ratio . in this model a flattening of the diffusion coefficient below 4 gv is required to match the b / c ratio at low energies . to better match primaries ( @xmath27
, he ) in the dc model , we introduced a steeper injection spectrum below 20 gv ; such a break , however , has almost no effect on secondaries ( @xmath28 , @xmath29 ) .
the existence of a sharp upturn below a few gev / nucleon follows from snr shock acceleration theory @xcite ; this is a transition region between thermal and non - thermal particle populations in the shock .
our model does not require a sharp break ( 0.3 in index is enough ) .
we use the dc model to calculate the lis spectra of protons and antiprotons and then use the drift model to determine their modulated spectra and ratio over the solar cycles with positive ( @xmath30 ) and negative ( @xmath31 ) polarity ( figs .
[ fig : prot ] , [ fig : pbar ] ) . the variations shown
depend on the tilt angle.wso/ ] this allows us to estimate the near - earth spectra for arbitrary epochs in the past as well as make some predictions for the future .
it may be also used to test the theory of heliospheric modulation @xcite .
it appears quite difficult to get agreement with b / c , p & he spectra , and antiprotons spectrum simultaneously in the framework of simple `` physical '' reacceleration / convection models ; this conclusion is mainly the result of the increased precision of the cr experimental data , and also the improved reliability of the calculations . what could be the origin of this failure , apart from the propagation models ? for b / c
, it seems unlikely that the cross - sections are sigificantly in error since for b production from c and o they are well measured at the relevant energies . for @xmath27 and he the gev bump produced by reacceleration seems to be outside the limits allowed by the modulation models and the observed fluxes , although the uncertainties in modulation are still considerable .
such a bump in the spectra of primaries ( @xmath27 , he , @xmath32 ) could be `` removed '' by including a flattening of the injection spectrum at low energies , while for secondaries ( @xmath29 , @xmath28 ) we have no such freedom and the problem remains .
if we assume then that the problem is in the propagation models , we have shown that it _ is _ possible to construct a model ( dc ) which fits all these data , by postulating a significant flattening of the diffusion coefficient below 4 gv together with convection ( and possibly with reduced reacceleration ) .
this type of break in the diffusion coefficient is reminiscent of the standard procedure in `` leaky - box '' models where the escape time is set to a constant below a few gev .
this has always appeared a completely _ ad hoc _ device without physical justification , but the present analysis suggest it may be forced on us by these type of data , so that possibilities for its physical origin should be studied . at the same time
the sensitivity of the whole analysis to the modulation models has to be investigated further .
we thank frank jones and vladimir ptuskin for fruitful discussions , and don ellison for discussions and a copy of his manuscript prior to publication .
ivm acknowledges support from nas / nrc research associateship program . | high energy collisions of cosmic ray ( cr ) nuclei with interstellar gas are believed to be the mechanism producing the majority of cr antiprotons .
the distinguishing spectral shape with a maximum at 2 gev and a sharp decrease towards lower energies makes antiprotons a unique probe of the models of particle propagation in the galaxy and modulation in the heliosphere .
besides , accurate calculation of the secondary antiproton flux provides a `` background '' for searches for exotic signals from the annihilation of supersymmetric particles and primordial black hole evaporation .
recently new data with large statistics on the antiproton flux have become available which allow for such tests to be performed .
we use our 3d galactic cosmic ray propagation code galprop to calculate interstellar propagation in several models . for our best model
we make predictions of proton and antiproton fluxes near the earth for different modulation levels and polarity using a steady - state drift model for heliospheric modulation . |
we shall use seiberg witten invariants to show that certain smooth structures obstruct the existence of einstein metrics , and refer the reader to @xcite for the definitions and basic properties of the invariants .
all manifolds in this section are closed , smooth , oriented @xmath0manifolds . for the sake of simplicity
, we assume @xmath6 throughout , though this is not essential . we shall need the following result concerning the behaviour of the invariants under connected summing with @xmath7 .
this is usually referred to as a blowup formula .
[ p : blowup ] let @xmath8 be a @xmath9structure on @xmath10 , and @xmath11 , with @xmath12 a generator of @xmath13
. then @xmath1 has a @xmath9structure @xmath14 with @xmath15 , such that the seiberg witten invariants of @xmath8 and of @xmath14 are equal ( up to sign ) .
as the reflection in @xmath16 in the cohomology of @xmath1 is realised by a self - diffeomorphism , the naturality of the invariants shows that there is another @xmath9structure with the same seiberg
witten invariant , up to sign , and with@xmath17 .
using this , we can prove the following version of a theorem of lebrun @xcite : [ t : blowup ] let @xmath10 be a manifold with a non - zero seiberg
witten invariant ( of any degree ) , and @xmath18 .
if @xmath19 , then @xmath1 does not admit an einstein metric . if @xmath14 has a non - zero seiberg
witten invariant , then for every riemannian metric @xmath20 there must be a solution @xmath21 of the monopole equations . denoting by @xmath22 the connection induced by the @xmath9connection @xmath23 on the determinant bundle of the spinor bundle ,
we have @xmath24 where @xmath25 denotes the scalar curvature of @xmath20 . given any class @xmath26 , denote by @xmath27 the projection of @xmath28 into the subspace @xmath29 of @xmath20self - dual harmonic forms along the subspace @xmath30 of @xmath20anti - self - dual harmonic forms .
the argument above really proves @xmath31 if @xmath8 is a @xmath9structure on @xmath10 with non - zero seiberg
witten invariant , then , by proposition [ p : blowup ] and the subsequent remark , there are @xmath9structures @xmath14 on @xmath18 with non - zero seiberg witten invariants and with @xmath32 for any choice of the signs @xmath33
. choose the signs so that @xmath34 then @xmath35 where we have used the inequality @xmath36 which is equivalent to the assertion that the moduli space associated with @xmath8 has non - negative dimension .
thus , we have proved @xmath37 for every metric @xmath20 on @xmath1 .
suppose now that @xmath20 is einstein .
then the chern
weil integrals for the euler characteristic and the signature of @xmath1 give @xmath38 where @xmath39 denotes the self - dual part of the weyl tensor of @xmath20 .
therefore @xmath40 , which implies @xmath41 .
theorem [ t : blowup ] was proved by lebrun @xcite , who also discussed the borderline case @xmath42 , in the case where @xmath10 is complex or symplectic . in that case
the blown up manifold @xmath1 is also complex , respectively symplectic , so that proposition [ p : blowup ] is not needed .
the following is the main result of this section , giving the examples mentioned in the introduction .
[ t : main ] there are infinitely many pairs @xmath43 of simply connected closed oriented smooth @xmath0manifolds such that : * @xmath44 is homeomorphic to @xmath45 , * if @xmath46 , then @xmath44 and @xmath47 are not homotopy equivalent , * @xmath45 admits an einstein metric but @xmath44 does not , * @xmath48 .
note that 3 ) implies in particular that @xmath44 and @xmath45 are not diffeomorphic .
we claim that there are simply connected minimal complex surfaces @xmath49 , @xmath45 of general type such that if we take @xmath50 , for a suitable @xmath51 with @xmath52 , then the pairs @xmath43 have all the desired properties .
the last property , the strict hitchin
thorpe inequality , follows from the noether and miyaoka yau inequalities for @xmath45 , which , by the first property , has the same euler characteristic and signature as @xmath44 . if we take @xmath45 to have ample canonical bundle , then the results of aubin and yau on the calabi conjecture show that @xmath45 admits a khler
einstein metric , compare @xcite . on the other hand , @xmath44 does not admit any einstein metric by theorem [ t : blowup ] .
the crucial issue then is to arrange that @xmath45 , with ample canonical bundle , is homeomorphic to the @xmath51fold blowup of @xmath49 , with @xmath53 .
as @xmath44 will be automatically non - spin , @xmath44 and @xmath45 will be homeomorphic by freedman s classification @xcite as soon as @xmath45 is non - spin and has the same euler characteristic and the same signature as @xmath44 .
one can find suitable surfaces using the known results on the geography of surfaces of general type , see @xcite for a summary of the results .
to exhibit concrete examples , instead of working with the topological euler characteristic and the signature , we shall use the first chern number @xmath54 and the euler characteristic of the structure sheaf @xmath55 . under blowing up
, @xmath56 drops by one and @xmath57 is constant .
thus , the miyaoka yau inequality for @xmath49 implies @xmath58 .
in fact , @xmath59 will be smaller still , because simply connected surfaces @xmath49 are not known to exist if we get too close to the miyaoka yau line @xmath60 . the minimal surface @xmath45 satisfies the same inequalities on its characteristic numbers as @xmath44 .
if the canonical bundle of @xmath45 is very ample , then castelnuovo s theorem , see @xcite page 228 , gives @xmath61 , which will contradict the above upper bound for @xmath59 .
thus , @xmath45 must be chosen to have ample but not very ample canonical bundle , and will be in the sector where @xmath62 the first being the noether inequality .
the results of xiao gang and z chen , cf @xcite , show that all non - spin simply connected surfaces @xmath45 with ample canonical bundle which are in this sector , and not too close to the line @xmath63 , will have companions @xmath44 as required , obtained by blowing up minimal surfaces @xmath49 . note that by beauville s theorem on the canonical map , cf @xcite page 228 , all the @xmath45 will be double covers of ruled surfaces .
we can avoid using the results of xiao and chen by taking for @xmath45 the following family of horikawa surfaces , cf @xcite .
let @xmath64 be the hirzebruch surface whose section at infinity @xmath65 has self - intersection @xmath66 , and let @xmath45 be a double cover of @xmath64 branched in a smooth curve homologous to @xmath67 , where @xmath68 is the class of the fiber .
the double cover is simply connected as @xmath69 is ample , and @xmath70 is not @xmath71divisible and so @xmath45 is not spin . moreover , @xmath72 is the pullback of an ample line bundle and therefore ample , so that @xmath45 admits a khler
einstein metric .
the characteristic numbers of @xmath45 are @xmath73 and @xmath74 .
now , by the classical geography results of persson @xcite , for all @xmath75 large enough there are simply connected surfaces @xmath49 of general type with @xmath76 and @xmath77 , so that the @xmath78fold blowup @xmath44 of @xmath49 is homeomorphic to @xmath45 . the pairs @xmath43 have all the desired properties .
the examples of manifolds without einstein metrics given by lebrun @xcite , namely blowups of hypersurfaces in @xmath79 , can not be used to prove theorem [ t : main ] because they violate the noether inequality .
they are therefore not homeomorphic to minimal surfaces for which the resolution of the calabi conjecture gives existence of an einstein metric . in view of theorem [ t : main ]
, one can ask how many smooth structures with einstein metrics and how many without , a given topological manifold has . on the one hand , using for example the work of fintushel stern , one can show that one has infinitely many choices for the smooth structures of the manifolds @xmath49 in the proof of theorem [ t : main ] , which remain distinct under blowing up points .
thus , one has infinitely many smooth manifolds one can use for each @xmath44 , not admitting any einstein metrics . on the other hand
, it is known that there are homeomorphic non - diffeomorphic minimal surfaces of general type , cf @xcite , page 410 , and the references cited there .
in fact , the number of distinct smooth structures among sets of homeomorphic minimal surfaces of general type can be arbitrarily large @xcite .
it is not hard to check that all the examples in @xcite and @xcite have ample canonical bundle , and therefore have khler einstein metrics of negative scalar curvature .
however , all those examples have @xmath80 , and can therefore not be used as the @xmath45 in the proof of theorem [ t : main ] .
still , those examples show that a given simply connected topological manifold can have an arbitrarily large number of smooth structures admitting einstein metrics .
compare theorem [ t : top ] below .
we have seen that existence of einstein metrics on closed @xmath0manifolds depends in an essential way on the smooth structure .
i believe that the issue of uniqueness , up to the sign of the scalar curvature , is also tied to the smooth structure . more specifically : such questions were raised in @xcite , pages 1819 , and are also addressed in @xcite .
what is new here , and in @xcite , is that the answer depends on the smooth structure , and also seems to depend on the dimension .
the conjecture is interesting because it is sharp it would be false if one did not fix the smooth structure , but only the underlying topological manifold : [ t : top ] there are simply connected homeomorphic but non - diffeomorphic smooth @xmath0manifolds @xmath1 and @xmath10 , such that @xmath1 admits an einstein metric of positive scalar curvature , and @xmath10 admits an einstein metric of negative scalar curvature . for @xmath10
we take the simply connected numerical godeaux surface constructed by craighero
gattazzo @xcite and studied recently by dolgachev
werner @xcite .
this has ample canonical bundle , so that by the work of aubin and yau it admits a khler
einstein metric of negative scalar curvature . by freedman
s classification @xcite , @xmath1 and @xmath10 are homeomorphic .
that they are not diffeomorphic is clear from @xcite .
the argument carried out there for the barlow surface , cf @xcite , works even more easily for the craighero
gattazzo surface as there is no complication arising from @xmath83curves .
alternatively , the fact that @xmath1 and @xmath10 have khler einstein metrics of opposite signs implies via seiberg
witten theory that they are non - diffeomorphic . in higher dimensions ,
homeomorphic non - diffeomorphic manifolds with einstein metrics are known @xcite , though in those examples all the metrics have positive scalar curvature .
although the examples of @xcite are consistent with a higher - dimensional analogue of the above conjecture , such a generalisation is false : let @xmath1 and @xmath10 be as in theorem [ t : top ] . then the @xmath75fold products @xmath86 and @xmath87 have khler einstein metrics of positive , respectively negative , scalar curvature .
however , as @xmath1 and @xmath10 are simply connected and homeomorphic , they are @xmath88cobordant .
therefore , @xmath89 and @xmath90 are also @xmath88cobordant for all @xmath75 , and for @xmath84 are diffeomorphic by the @xmath88cobordism theorem .
theorem [ t : top ] and corollary 2.3 have also been proved independently by catanese and lebrun @xcite . instead of the craighero
gattazzo surface they use the barlow surface , showing that it has deformations with ample canonical bundle .
conjecturally , the barlow and craighero
gattazzo surfaces are deformation equivalent , and therefore diffeomorphic . | we prove that there are infinitely many pairs of homeomorphic non - diffeomorphic smooth @xmath0manifolds , such that in each pair one manifold admits an einstein metric and the other does not .
we also show that there are closed @xmath0manifolds with two smooth structures which admit einstein metrics with opposite signs of the scalar curvature .
gtoutput in dimensions strictly smaller than four einstein metrics have constant curvature and are therefore rare . in dimension four
einstein metrics of non - constant curvature exist , but it is still the case that existence of such a metric imposes non - trivial restrictions on the underlying manifold . for closed orientable einstein @xmath0manifolds
@xmath1 the euler characteristic has to be non - negative , and , furthermore , the hitchin
thorpe inequality @xmath2 must hold , where @xmath3 denotes the euler characteristic and @xmath4 the signature .
this condition is very crude , and is certainly homotopy invariant , as are the restrictions coming from gromov s notion of simplicial volume @xcite , and from the existence of maps of non - zero degree to hyperbolic manifolds @xcite .
our aim in this note is to discuss existence and non - existence of einstein metrics as a property of the smooth structure .
we shall exhibit infinitely many pairs of homeomorphic non - diffeomorphic smooth @xmath0manifolds , such that in each pair one manifold admits an einstein metric and the other does not .
this shows for the first time that the smooth structures of @xmath0manifolds form definite obstructions to the existence of an einstein metric .
an isolated example of such a pair can be obtained as follows .
hitchin @xcite showed that einstein manifolds for which is an equality are either flat or quotients of a @xmath5 surface with a calabi
yau metric .
thus , the existence of smooth manifolds homeomorphic but not diffeomorphic to the @xmath5 surface , which is known from donaldson theory @xcite and also follows easily from seiberg
witten theory , see for example @xcite , shows that by changing only the differentiable structure one can pass from a manifold with an einstein metric to one without .
the point of our examples is that there are lots of them , and they do not arise from the borderline case of a non - existence result .
they are in some sense generic .
we shall also discuss a conjecture concerning uniqueness of einstein metrics on @xmath0manifolds which complements the discussion of existence .
this too depends on a consideration of different smooth structures on a fixed topological manifold . |
there is a general consensus that surface defects such as steps , kinks , corners and surface anion and cation vacancies play a crucial role in molecular processes at oxide surfaces @xcite . yet in a very general sense the background understanding needed to discuss these processes is seriously lacking .
there is very little direct experimental information about the structure and energetics of these surface defects for most oxides ( see , for instance , references @xcite ) .
it is true that there has been useful theoretical work on the relaxed structure of steps and other surface irregularities ( see e.g. references @xcite ) based on simple ( and usually empirical ) interaction models , but the validity of the assumptions on which the models are based is not always clear .
important insights have also been gained _ via _ semi - empirical quantum calculations @xcite . in a few cases ,
fully _ ab initio _
calculations on oxide surfaces have been reported @xcite , but these are almost all restricted to flat non - defective surfaces ( see , however , ref .
@xcite ) .
the aim of the present paper is to contribute to the background understanding of defective oxide surfaces by reporting a series of _ ab initio _ total - energy calculations on the relaxed atomic structure and the electronic structure of the flat mgo ( 001 ) surface , of steps , corners and reverse corners ( see section 5 for the complete explanation of these terms ) on this surface , and of surface f - centers ( two electrons trapped in an oxygen vacancy ) at the flat surface and at the step , corner and reverse corner . for reference purposes , we also present results on the bulk f - center .
our particular interest is in the processes of oxygen exchange between the gas phase and the crystal lattice , and the present work is intended to provide some of the essential information needed for describing these processes , but we believe our results will also be relevant to other problems , such as the interpretation of the spectroscopic properties of the mgo surface .
our calculations are based on the density - functional pseudopotential approach @xcite which has already proved highly successful in many previous studies of oxide surfaces @xcite and point defects in oxides @xcite . in order to treat defective and irregular oxide surfaces
, one needs to do large numbers of calculations on systems containing up to ca .
60 atoms with full relaxation of the system to mechanical equilibrium . with the _ ab initio _
techniques used in the present work , this can now be done fairly routinely .
the paper is organized as follows . in the following section ,
we summarize the techniques we have used for calculating the ground - state total energy of the system and the electronic density of states .
we then present our results on the bulk f - center ( section 3 ) .
this is followed by a description of our results for the flat ( 001 ) surface with and without f - centers ( section 4 ) and for the step , corner and reverse corner and for f - centers at these features ( section 5 ) .
a discussion of our results is given in section 6 .
density - functional theory and the pseudopotential approximation are well established and widely used methods , which have been extensively reviewed in the literature ( see e.g. references @xcite ) .
only valence electrons are represented explicitly in the calculations , the valence - core interaction being described by non - local pseudopotentials which are generated by _ ab initio _
calculations on isolated atoms .
the effects of electron correlation are included using the local density approximation ( lda ) . a large amount of work on mgo based on this approach
has already been reported @xcite .
the calculations are performed on periodically repeated cells , with every occupied valence orbital represented as a plane - wave expansion .
this expansion includes all plane waves whose kinetic energy @xmath0 ( * k * the wavevector , _ m _ the electronic mass ) satisfies @xmath1 , where @xmath2 is a chosen plane - wave cut - off energy .
this means that calculations can be taken to convergence with respect to the size of basis set simply by increasing the cut - off . in the present work
, the self - consistent ground state of the system is determined using the car - parrinello approach @xcite , in which the total energy is minimized with respect to the plane wave coefficients of the occupied orbitals , the minimization being performed by the conjugate - gradient technique @xcite .
the calculations were performed with the code cetep running on the cray t3d parallel supercomputer at the edinburgh parallel computer centre .
the computational strategy underlying the cetep code has been described in the literature @xcite , but the code has been extensively rewritten to take full advantage of the t3d .
the main change concerns the way the conjugate gradient search is done . in previous versions , each band was updated sequentially in the iterative search for the ground state . in the version used here ,
all occupied bands are updated simultaneously .
this method exploits the large memory available on the t3d , and results in a speed - up of about a factor of three for the present type of problem .
technical details of the calculations are as follows .
the pseudopotentials for mg and o atoms are identical to those used in our recent calculations on the adsorption of nh@xmath3 at the mgo ( 001 ) surface @xcite . as in that work
, the kleinman - bylander representation @xcite of the pseudopotentials is used , with the _ s _ component treated as local for mg and the _ d _
component for o. the only difference with this work lies in using the real - space representation @xcite of the pseudopotentials , which speeds up the calculations significantly for systems of the large size treated here .
we demonstrated in our earlier work @xcite that with a plane - wave cut - off @xmath2 = 600 ev the total energy is essentially perfectly converged with respect to the size of the basis set , and the same cut - off has been used here .
electronic exchange and correlation are represented by the commonly used ceperley - alder form @xcite .
brillouin zone sampling for the ground state calculations is performed using the monkhorst - pack scheme @xcite , as noted in more detail later .
it was found in our previous calculations @xcite that in equilibrium the nearest - neighbor distance is @xmath4 , and we have used the same value in our present calculations .
note that it is close to the experimental value 2.105 @xcite . for all systems studied in this paper
we have calculated the electronic density of states ( dos ) @xmath5 where @xmath6 is the eigenvalue of the kohn - sham equation for the band @xmath7 and the wave vector * k * in the brillouin zone ( bz ) of the supercell , _
n _ is the number of unit cells in the crystal . when making calculations on _ bulk _ systems , the standard tetrahedron method has been used @xcite
first , we note that the bz can be chosen as a parallelepiped defined by the three primitive reciprocal lattice vectors @xmath8 , @xmath9 , @xmath10 .
then , we divide the bz into small parallelepipeds of an equal volume by dividing every lattice vector @xmath11 ( @xmath121,2,3 ) by integers @xmath13 , @xmath14 , @xmath15 .
every small parallelepiped is then split into six tetrahedra of an equal volume @xcite , and all the * k*-points at the edges of the tetrahedra form a submesh of * k*-points .
using the point - group symmetry operations of the system , a significantly reduced set @xmath16 of symmetry inequivalent * k*-points is obtained , and an array containing the correspondence between these * k*-points and all edges of all the tetrahedra is generated .
then , using the castep code @xcite , we calculate the eigenvalues @xmath6 for every * k*-point belonging to the set @xmath16 . using the array which sets up the correspondence between the set @xmath16 and all edges of tetrahedra ,
the total dos is finally calculated .
several points are worth mentioning .
first , we emphasize that for systems of the size treated here the calculations of the eigenvalues @xmath6 are extremely demanding and we had to restrict our * k*-point sampling by choosing equal divisions @xmath17 corresponding to 750 tetrahedra in the whole bz .
for high - symmetry systems without an f - center ( except for the perfect corner - reverse corner system which has no symmetry at all ) we have used also a larger number of divisions to study the convergence of the total dos .
we found this division to be reasonable and reliable as far as the boundaries and the general structure of bands are concerned .
however , we can not rely on the fine structure of the total dos since the energy resolution appears to be not completely satisfactory .
second , when making dos calculations for the surface systems in slab geometry , it is unnecessary to consider the * k*-point sampling along the direction of the normal to the surface @xcite , so that the bz is reduced to the two - dimensional _ surface _ bz . in this case , the method of tetrahedron is simplified to the triangle method .
third , since the * k*-point sampling was not very extensive in all our calculations , we found that often there is an accidental equivalence of some edges of one and the same tetrahedron which results in an accidental degeneracy of the eigenvalues , @xmath6 .
this forced us to modify slightly the standard method of tetrahedra to cope with this problem .
namely , we have formally removed this degeneracy by adding different @xmath18 to the eigenvalues belonging to the equivalent edges .
then , using equations of the standard method , the corresponding modified formulae have been worked out by making the limit @xmath19 .
we have used the techniques described in the previous section to perform calculations on the bulk f - center using repeating cells containing 8 , 16 , 32 , 54 and 64 lattice sites . these are all the repeating cells containing between 8 and 64 sites , which preserve the o@xmath20 symmetry of the f - center , and they give periodic arrays of f - centers having s.c . , f.c.c .
, f.c.c . and s.c .
translational symmetry , respectively .
the next available size of the cell would contain 128 sites . for each repeating cell
we have first calculated the total energy of the perfect crystal .
we have then removed an oxygen atom and repeated the calculation .
in addition to treating the unrelaxed f - center , we have also studied its equilibrium by relaxing all ions in the system .
all calculations have been performed using the lowest - order monkhorst - pack set of * k*-points , consisting of the eight points @xmath21 ( in units of the primitive reciprocal lattice vectors ) . as usual , it is unnecessary to include @xmath22-points related by inversion symmetry , so that the calculations are actually performed with the four points @xmath23 .
we first report our results for the oxygen removal energy .
this is defined to be the energy ( per cell ) of the system containing the f - center , plus the energy of an isolated oxygen atom minus the energy of the perfect crystal calculated with exactly the same cell . in view of the large energies involved ,
it is important that the same cell be used for the f - center and perfect crystal to ensure cancellation of errors .
the energy of the isolated oxygen atom was obtained by calculations on a number of periodic systems containing one oxygen atom per cell .
the final size of the repeating cell in these calculations has been taken large enough to ensure that the interaction between oxygen atoms introduces an error of less than 0.01 ev . since an oxygen atom in the ground state
has a multiplet structure @xmath24p@xmath25 @xcite , and the state calculated by the lda method belongs to the excited state @xmath26d of the atom , a correction has to be made for the energy of the oxygen atom . as an estimate for this correction , we have used the experimental energy difference for these states @xcite which is 1.967 ev . note that the local spin density ( lsd ) calculations give a somewhat similar correction ( the difference between energies calculated using lsd and ld methods ) which is 1.496 ev @xcite and 1.4 ev @xcite . to estimate the overall error in our oxygen removal energies ,
we have also calculated the cohesive energy ( per formula unit with respect to free neutral atoms ) of the mgo crystal .
using the total energy of the largest repeating cell for the bulk containing 64 sites and the calculated total energies of the o and mg atoms , we have calculated the cohesive energy to be 11.14 ev .
note that we have also used the 0.14 ev correction @xcite for the zero - point vibrational energy of mgo .
our value turns out to overestimate the experimental cohesive energy ( 10.3 ev @xcite ) .
the existence of this discrepancy with experiment of 0.8 ev suggests that there may be a similar systematic error in calculated oxygen removal energies .
our calculated oxygen removal energies for both the unrelaxed and relaxed f - center are reported in table 1 .
two main conclusions are clear from these results .
first , the removal energies are completely converged with respect to the size of the cell already for the 16-site system .
in fact , even the 8-site result differs from the others by only 0.1 ev .
second , the effect of relaxation is completely negligible .
the almost complete absence of relaxation effects for the neutral f - center in mgo has also been demonstrated by earlier calculations @xcite .
the comparison of our calculated oxygen removal energy with experiment and with other calculations will be discussed later .
although the relaxations are extremely small , they do show systematic behaviour for all the sizes of cell we have studied .
the mg ions adjacent to the f - center relax outwards by an amount which is 0.01 for the 54-site and 64-site cells and is somewhat smaller for the smaller cells .
the oxygens nearest to the f - center relax outwards by about 0.005 for the two largest cells .
the relaxations surrounding the f - center in mgo have been studied previously by _ ab initio _
calculations @xcite .
all these calculations agree that the displacements are extremely small and of the same general magnitude that we have found , but there are some disagreements about the directions of the displacements .
these disagreements can not be considered very significant , given the very small effects involved .
we have calculated the electronic dos for the perfect crystal and for the system containing an f - center , using the tetrahedron technique described in the previous section .
it is straightforward to use enough @xmath22-points to obtain good energy resolution for the perfect crystal , but the calculations on large cells are much more demanding , and the energy resolution for the f - center systems is less good .
we show in fig .
1 our results for the dos of the perfect crystal and for the 16-site and 54-site f - center systems . as is well known from many previous calculations , the valence dos of the perfect crystal consists of a rather narrow band of o(2_s _ ) states , and a broader two - peak band of o(2_p _ ) states .
we have found in our calculations that the widths of these bands are 2.1 ev and 4.9 ev , respectively , and we estimate the separation between two peaks in the band of o(2_p _ ) states to be ca .
our calculated separation between bands of o(2_s _ ) and o(2_p _ ) states is ca .
the widths , positions and shapes of the valence bands in the dos are in close agreement with previous theoretical results @xcite .
agreement with experiment @xcite ( 2.0 - 2.5 ev and 5.0 - 7.0 ev for the widths of the two bands and 13 - 14.5 ev for the _ s - p_-bands separation ) is somewhat worse but this is usual with lda calculations .
in addition , the bulk band gap between the top of the upper valence band and the bottom of the conduction band is underestimated .
our calculated value is 4.8 ev , which should be compared with the experimental value of 7.8 @xcite ( note that a bigger value 8.7 ev is reported in @xcite ) .
this degree of underestimation of band gaps is very typical for lda calculations .
the states associated with the f - center are clearly visible in the lower half of the band gap in fig .
1 . an isolated f - center would have a single sharply defined level .
in the periodic systems treated here , this level is broadened into a band . as expected , this band is broader for the smaller system .
the width of this band is 1.6 ev and 0.6 ev for the 16-site and 54-site systems .
the large width for the small system suggests that the f - center wavefunction is rather extended , and we shall analyze this question below .
the average energy ( i.e. the first moment ) of the f - center band is ca .
2.5 ev above the valence band maximum ( vbm ) for the 16-site system and 2.7 ev above the vbm for the 54-site system .
this can be compared with the position deduced from optical absorption measurements .
the observed f - center absorption energy is ca .
5.0 ev @xcite and it is generally agreed @xcite that this represents a transition from the ground state to a state lying just below the conduction band .
taking the band gap to be 7.8 ev @xcite , we deduce that the f - center level should be roughly 2.8 ev above the vbm .
our calculated position of the f - center level is also in close agreement with values obtained in other _
ab initio _ calculations ( 2.34 ev @xcite , 2.55 ev @xcite ) . a contour plot of the valence electron density in the 54-site f - center system is shown in fig .
as expected from earlier work @xcite , the density has a maximum at the center of the vacancy , but the electron distribution in the vacancy region is rather broad . to understand this in more detail , it is useful to compare the electron distributions in the f - center system and the perfect crystal . we have done this by calculating for each system the total valence charge inside spheres of different radii @xmath27 centered at the vacancy .
we denote this total charge by @xmath28 for the f - center system and @xmath29 for the perfect crystal .
we are interested in finding out how rapidly the disturbance due to the f - center dies away with increasing radius .
we therefore subtract @xmath29 from @xmath28 . to allow for the fact that there are six more electrons in the perfect crystal , it is convenient to study the quantity @xmath30 with this definition
, @xmath31 should tend to zero for large @xmath27 .
our results for @xmath31 are shown in fig .
these results show that the electron distribution outside a radius of approximately 2.0 is essentially the same in the f - center system as in the perfect crystal .
bearing in mind that the distance to the nearest mg ion is 2.08 , this rather good localization of the charge associated with the f - center explains the almost complete absence of relaxation effects .
we have also investigated the electron density associated with the f - center band itself . a contour plot of this density is shown in fig .
it is clear from this plot that the density of the f - center band is not completely localized in the vacancy , but has considerable weight on the nearest and even next - nearest oxygen ions .
we have analyzed this quantitatively and found that there is roughly one electron in the vacancy , 0.3 electron on nearest oxygens and the remaining 0.7 electron is distributed in weak features in a more distant regions .
this indicates that the electron density on oxygen neighbors is associated with states both at the valence band energies and also at the energy of the f - center level .
conversely , the electron density in the vacancy is associated with states both at the energy of the f - center level and also at the valence - band energies .
these results are in accord with the calculations of wang and holzwarth @xcite .
before describing our calculations on the surface f - center , we present briefly our results for the perfect surface . in all our surface calculations ,
we use slab geometry , so that the system we study is a periodically repeated stack of slabs separated by vacuum layers . for the calculation on the perfect surface ,
we use a vacuum width of 4.164 , which our previous work @xcite on the mgo ( 001 ) surface showed to be enough to make the interaction between slabs negligible . in the present calculations , we have studied slabs containing 2 , 4 and 6 layers with 16 , 32 and 48 ions respectively in the surface unit cell , and we found that the properties of interest are essentially the same in all three cases .
we have performed @xmath22-point sampling using the lowest - order monkhorst - pack scheme . since sampling along the surface
normal is unnecessary @xcite , this means that we use the two * k*-points @xmath32 where the third component is along the surface normal .
there have been a number of experimental studies @xcite and previous calculations @xcite on the properties of the ( 001 ) surface , which indicate that relaxation effects are extremely small .
this is fully confirmed in the present calculations . in accord with all previous work
, we find that surface oxygens relax outwards and surface magnesiums relax inwards .
for the six - layer slab , the values of these displacements are @xmath33 and @xmath34 .
these values are in qualitative agreement with those obtained in our recent pseudopotential lda calculations @xcite using a three - layer slab , which gave @xmath35 and @xmath36 .
the total electronic dos for the unrelaxed six - layer slab is shown on the top panel of fig .
the most striking feature is the marked reduction of the band gap .
this has a value of 3.2 ev , which is 1.6 ev smaller than our calculated value for the perfect crystal .
the figure indicates that the gap states are pulled down from the conduction band , rather than splitting from the top of the valence band .
our results concerning the reduction of the band gap as one goes from the bulk to the surface are in good agreement with the measurements by energy - dependent electron - energy - loss spectroscopy ( eels ) @xcite where a 2 ev reduction of the band gap has been observed . a reduction of the band gap has been also obtained in a number of previous calculations @xcite though the calculated values differ .
for example , surface states below the bottom of the conduction states were found in recent _ ab initio _
calculations @xcite , which gave the reduction of the band gap by ca .
we shall return to the discussion about the surface states in section 6 . in some oxides
, it has been found @xcite that the change of the madelung potential at the surface causes the o(2_s _ ) states of surface oxygens to split off from the top of the o(2_s _ ) band .
we do not observe this in the present case .
this is because the madelung potential is almost unchanged at surface oxygen sites ( the madelung constant at a ( 001 ) surface site in the rock - salt structure is 1.682 @xcite , while the value at a bulk site is 1.748 ) and that is consistent with the absence of oxygen - based surface states .
we have investigated the nature of the surface states by calculating the fictitious charge density associated with several of the lowest unoccupied bands . in practice
, we have calculated the quantity @xmath37 defined by @xmath38 where @xmath39 is the eigenfunction of band @xmath40 at wavevector @xmath22 from the irreducible wedge of the bz , and @xmath41 is the weighting factor for this wavevector .
the quantity @xmath37 represents the electron density that would be associated with orbital @xmath40 if the orbital were occupied .
shows @xmath37 for the lowest unoccupied band , i.e. the band immediately above the top of the gap .
the localization of @xmath37 in the surface region confirms that this band consists of surface states .
in contrast to the results of another _ ab initio _ lda calculation @xcite where surface states were found to be predominantly of mg _ s _ character , our surface states are concentrated mainly on surface oxygens and above surface magnesiums .
the anisotropic nature of @xmath37 around the mg ions can be understood in tight - binding terms as arising from the mixing of mg(3s ) and mg(3p ) states because of the electric field at the surface .
this mixing will also cause a lowering of one of the 3s/3p hybrid states , and this is one way of understanding the appearance of gap states below the conduction band minimum ( cf .
discussion in references @xcite ) .
we have also examined the valence electron density in the surface region , but we do not show it here , since the surface distribution will become clear in the next section .
we have investigated the properties of surface f - centers using calculations on 2-layer , 4-layer and 6-layer slabs .
the same surface unit cell was used in all cases .
the translation vectors joining nearest neighbor f - centers are in the ( 110 ) direction , and their length is @xmath42 , where @xmath43 is the cation - anion separation in the perfect crystal . with this surface unit cell ,
the repeating unit contains 16 , 32 and 48 sites in the three slabs . for the 4-layer slab ,
we have studied f - centers both in the surface layer and in the layer immediately below the surface , and for the 6-layer slab , we have also studied the f - center in the third layer .
each repeating unit contains only a single f - center , so that the slab contains an f - center only at one of the surfaces .
the vacuum width and the @xmath22-point sampling are the same as for the calculations on the perfect surface . in all cases
we have investigated both the unrelaxed and the fully relaxed systems .
we report in table 2 two oxygen removal energies for each case .
the first removal energy refers to the energy required to extract an oxygen atom from the slab with all atoms held fixed at their perfect lattice positions .
the second removal energy refers to the same process except that both the initial and the final systems are fully relaxed .
it is the second of these energies that is relevant to the real physical process .
four important conclusions emerge from these results .
the first is that relaxation effects are extremely small .
the largest difference between the two removal energies is only 0.06 ev , which is insignificant for practical purposes .
the second conclusion is that the removal energies are insensitive to the slab thickness . even for the 2-layer slab
the removal energy for the top layer differs by only 0.2 ev from the value for the 6-layer slab .
the differences between the 4-layer and 6-layer slabs are even smaller .
this indicates that the results are well converged with respect to the thickness of the slab .
the third conclusion is that the removal energies for the second and third layers are extremely close to the bulk value of 10.55 ev reported in table 1 .
however , the first - layer removal energy of 9.77 ev is substantially lower than the bulk value .
this effect is expected became of the 5-fold coordination of the surface oxygen atoms . as in the case of the bulk f - center ,
the ionic relaxations are very small . for the third - layer f - center ,
the relaxations are essentially the same as for the bulk f - center .
for the surface f - center , the neighboring mg ions relax away from the f - center in the plane of the surface by 0.02 , which is about twice the relaxation found in the bulk .
in addition , they have a component of displacement along the outward surface normal which is 0.04 with respect to their relaxed surface positions .
this implies a net outward relaxation of 0.02 of these neighboring mg ions with respect to their positions on the perfect surface .
the mg ion directly below the first - layer f - center relaxes away from the f - center by 0.05 .
the relaxation of neighboring oxygens is negligible .
the electronic dos for the systems containing first - layer and third - layer f - centers are shown on the bottom and middle panels in fig .
the dos for the second - layer f - center is almost the same as for the third - layer case .
the electronic level of the unrelaxed first - layer f - center is 2.0 ev above the top of the valence band , which becomes 2.3 ev on allowing the system to relax , so that relaxation effects are significant .
this f - center energy is slightly below that of the bulk f - center , which is 2.7 ev above the valence band maximum ( see section 3 ) .
so far as we are aware , there is no direct experimental evidence on this question .
however , it may be relevant that a downward shift of about 0.2 ev in the luminescence of surface f - centers ( with respect to that of bulk f - centers ) was observed for mgo in experiments on absorboluminescence @xcite . a slight downward shift of about 0.04 ev in the surface f - center absorption band in kcl crystal which is similar to mgo crystal studied here ,
was also reported in @xcite .
the shift of the f - center level in the direction of lower energies found in our calculations can be explained by a competition of a number of factors which work in opposite directions .
indeed , it was already mentioned above that the total coulomb ( hartree ) potential becomes weaker as one moves from the bulk to the surface . at the same time , the f - center wave function is more diffuse in this case as can be judged from our f - center electronic density distribution which is discussed below .
both these factors work in lifting up the f - center level . on the other hand ,
the positive kinetic energy of a more diffuse f - center state becomes smaller which results finally in a small downward shift of the f - center level .
for the second layer f - center the unrelaxed and relaxed values for the f - center level above the vbm are 2.5 and 2.56 ev , whereas for the third - layer f - center the value is 2.6 ev , which is very close to the bulk value .
it is worth commenting briefly that the band gap for the first - layer f - center system is slightly greater than for the perfect surface ( 3.5 ev rather than 3.2 ev ) .
such an effect can not occur for an isolated f - center , and must be due to the finite concentration of f - centers because of the periodic boundary conditions .
the valence electron density of the system containing the first - layer f - center is shown in fig .
the electronic density shows a maximum near the center of the vacancy , and it is interesting to note that this maximum value ( 0.13 electrons @xmath44 ) is only slightly below the corresponding maximum of the bulk f - center ( 0.15 electrons @xmath44 ) .
the electronic density decays rapidly along the outward surface normal .
the conclusion is that the f - center is highly localized in this 5-fold coordinated environment .
we have also examined the electron density of the surface f - center level itself , and we found that it is extremely similar to that of the bulk f - center .
the valence electron distributions for the systems containing second and third layer f - centers show no significant differences from the distributions for the bulk f - center .
our results show that already the third - layer f - center is very similar to the bulk f - center .
we have studied three kinds of surface irregularities : the step , the corner and the reverse corner , as shown in fig .
8 . as before
, the calculations are done using slab geometry , but the repeating unit is now somewhat more complicated . for the step calculations , we use a repeating cell containing 52 sites with each surface of the slab consisting of terraces of width @xmath45 , separated by steps of height @xmath43 .
the slab contains four layers of ions and the vacuum width is @xmath45 .
we expect this thickness of the slab to be adequate because the calculations described in the previous section show that the oxygen removal energy from the flat surface is essentially the same for the 4-layer and 6-layer slabs . in order to achieve this geometry for the step system , the translation vectors @xmath46 of the supercell are @xmath47 the orientation of the coordinate system corresponding to
this choice of the lattice vectors is clear from fig .
( it will be noted that for the perfect step , the vector @xmath48 could be equally well be taken to be @xmath49 ; the reason for our choice of @xmath48 will be explained below . )
the corner and reverse corner systems are studied using a repeating cell of 44 sites .
each surface of the slab consists again of terraces and steps , but now the steps zig - zag back and forth along the ( 100 ) and ( 010 ) directions .
corners and reverse corners alternate , with the distance between each corner and the neighboring reverse corner being @xmath50 . the slab thickness and the vacuum width are the same as before .
the translation vectors of the supercell are @xmath51 the lowest order monkhorst - pack set of two @xmath22-points is used .
the relaxation energies of both the step and the corner systems are substantial . for the step system ,
the total relaxation energy per repeating cell is 1.51 ev , which corresponds to 0.38 ev per ion pair along the length of the step .
the total relaxation energy for the corner system is 3.01 ev per cell , or 0.75 ev per ion pair along the length of the zig - zag step .
the ionic displacements away from their perfect - lattice positions for the four inequivalent atoms at the step itself are reported in table 3 .
our results show that the displacements are quite substantial and their general direction is such as to smooth out the discontinuity at the step .
the ionic displacements associated with the corner and the reverse corner system are reported in table 4 .
once again , we see substantial displacements having the same general tendency as for the perfect step .
it is worth noting that the displacements of the mg ion number 4 at the bottom of the reverse corner and of its oxygen neighbors 5 are very small .
the step on mgo ( 001 ) surface has previously been studied both by empirical modeling @xcite and by semi - empirical quantum calculations @xcite , but we can not provide a detailed comparison since the displacements have not been reported in these earlier studies .
however , the magnitudes and directions of the displacements that can be inferred from the pictorial data given in @xcite seem to be in general agreement with those we have found .
the electronic dos for the relaxed step is almost identical to that of the flat ( 001 ) surface , and we do not show it here . just as for the flat surface , there is a band of surface states pulled down from the bottom of the conduction band .
the band gap for the step system is 3.3 ev , which is slightly larger than that for the flat surface ( 3.2 ev , see section 4.1 ) , but we do not regard this as significant .
thus , our results on the band gap for the step system do not support experimental suggestion @xcite of a substantial 0.85 ev lowering of the excitonic surface states associated with four - fold coordinated oxygens on the surface with respect to those for the flat ( 001 ) surface .
the only other noticeable effect is a downward shift of the o(2_p _ ) valence band with respect to the o(2_s _ ) band by about 0.2 ev . the dos of the corner - reverse corner system shown in fig .
9 is much more interesting .
first , there is a narrow band of states completely split off from the top of the o(2_s _ ) states by ca .
second , there is a significant peak at the top of the o(2_p _ ) valence band , though this is not split off .
third , there is a band of states split off from the bottom of the conduction band .
the gap between the top of the valence band and the lowest of the unoccupied states split off from the bottom of the conduction band is 2.4 ev , which is considerably less than that for the flat surface and the step .
the uv diffuse reflectance spectroscopy @xcite has also demonstrated almost 2.0 ev lowering of the surface excitonic states attributed to the three - fold coordinated surface oxygen atoms ( i.e. at the corners ) with respect to the band gap for the flat ( 001 ) surface .
previous cluster lmto calculations @xcite which found a substantial reduction of the gap at the corners compared with the flat surface of a cluster are also consistent with our calculations .
we think that the appearance of the surface states above the bands of o(2_s _ ) and o(2_p _ ) states can be simply explained by a drastic decrease of the hartree potential at the corner site ( as follows from @xcite ( see also @xcite , p. 117 ) , the madelung constant at the corner site is only 0.87378 , i.e. half the value for the five - coordinated surface site where it is 1.68155 ) .
the fact that the surface states are not split off completely from the top of the upper valence band is the result of a more delocalized character of these states in comparison with the o(2_s _ ) surface states . as in the case of the flat surface
, we have investigated the nature of the surface states by calculating the electron density associated with them , see eq.(3 ) . for the step system , we find that the surface states at the bottom of the conduction band are extremely similar to those at the flat surface .
for the corner - reverse corner system , the surface states split off from the o(2_s _ ) band have _ s _ character and are strongly localized on the corner oxygens with a substantial contribution from the three nearest oxygens on the terrace underneath .
the surface states at the top of the valence band have _ p _ character and are also localized mainly on the corner oxygens ; the axis of the _
p_-like density distribution passes through the corner site and lies almost in the ( 111 ) direction .
the states at the bottom of the conduction band are shown in fig .
10 , which indicates that they are strongly associated with mg ions neighboring the reverse corner , though there is also considerable weight on surface oxygen ions .
the total spatial densities of valence electrons in the perfect step and corner systems do not show very remarkable features .
however , in order to provide a point of reference for the f - center densities shown later , we display in fig .
11 the density associated with the corner system on a plane parallel to the terraces passing through the zig - zag step .
it is noteworthy that the large displacement of the corner oxygens leads to a significant build - up of charge between these ions and their oxygen neighbors .
we have calculated the energy for removing oxygen from the step , the corner and the reverse corner . in the case of the step
, we have positioned the f - centers so as to maximize the separation between them , as shown in fig .
8_a_. this is the reason for taking the translation vectors @xmath46 for the step as shown in section 5.1 ( see eq.(4 ) ) . in fig .
8_b _ , we show the location of the f - centers at the corner and the reverse corner . the unrelaxed and relaxed oxygen removal energies are given in table 5 .
we note the following points . at the reverse corner ,
the oxygen removal energy is only slightly less ( by 0.4 ev ) than its value for the first - layer f - center on the flat surface . in both these cases ,
the f - centers are 5-fold coordinated . a further reduction of 0.4 ev occurs when the f - center is formed at the 4-fold coordinated step site , and the removal energy decreases by 0.9 ev more when we go to the 3-fold coordinated corner site .
overall , the removal energy is 2.5 ev lower at the corner site than in the bulk .
the difference between unrelaxed and relaxed removal energies is very small except for the corner site , where it is a little over 1 ev .
the ionic displacements around the f - center for the step system are reported in table 3 , whereas those for systems with a f - center at the corner and reverse corner sites are given in table 4 .
although the general trend is the same as in all systems discussed so far , namely it is an additional outward movement of nearest mg ions from the f - center site , the additional displacements are found to be substantial only for the f - center at the step and at the corner .
the electron dos for the system containing the f - center at the step is shown in fig . 12 .
this is essentially identical to the dos for the perfect step system , except for the presence of the narrow f - center band in the gap .
the mean energy of this band lies 2.4 ev above the top of the valence band , which is almost identical to its position for the f - center on the flat surface .
the dos for the f - center at the corner and the reverse corner are shown in fig .
the simplest case to understand is that of the f - center at the reverse corner , for which the dos is very similar to that of the perfect corner - reverse corner system except for the presence of the f - center band in the gap , whose mean position lies only 0.8 ev above the top of the band of o(2_p _ ) surface states .
the dos of the corner f - center is more complicated , because in addition to the appearance of the f - center band , there also are substantial changes to the surface states .
the mean position of the f - center band is 1.7 ev above the top of the band of o(2_p _ ) surface states , which is considerably higher than that for the reverse corner case .
this can mainly be explained by the reduction of the hartree potential at the corner site and by the strong delocalization of the f - center wave function associated with the f - center band .
we have also looked at the spatial distribution of some interesting features in the dos .
we found that because of the absence of oxygen atoms at the corner sites on the upper side of the slab , the states associated with the band split off from the top of the o(2_s _ ) band have changed their character of localization .
namely , they appear to be more delocalized over a larger number of oxygens which are behind the `` empty '' corners on the same terrace . the localization of these states in the two lowest layers is not changed .
surprisingly , the energy position of this band is also not changed with respect to that for the perfect corner - reverse corner system .
in addition to that , we have found similar changes in the nature of states associated with the surface states at the top of the o(2_p _ ) band .
unoccupied states attributed to the band split off from the bottom of the conduction band also show considerable delocalization over the corner regions along the direction of the zig - zag , as well as over oxygen atoms which are behind the corner sites on the same terraces .
the electron densities associated with the f - center state at the step and the reverse corner are qualitatively rather similar to what we have already shown for the flat ( 001 ) surface . as in that case
, the density is strongly localized at the f - center site , with substantial weights on neighboring oxygens .
the height of the peak in the density at the reverse corner f - center site is almost identical to that found at the surface f - center , as might be expected since both sites have the same coordination .
the height of the peak at the step f - center site is about 30 percent lower . on the other hand ,
the density associated with the corner f - center state shows considerable delocalization as demonstrated in fig . 13
we first comment on the relaxed structure and the electronic structure of the mgo surface without f - centers .
our results for the flat ( 001 ) surface confirm what is already known from leed @xcite measurements , from previous _ ab initio _ @xcite and semiempirical quantum calculations @xcite , and from modelling based on interaction models @xcite that relaxation effects are extremely small , and consist mainly of rumpling , with o ions moving slightly out of the surface and mg ions moving slightly in .
however , we have shown that relaxation effects are far larger at the step , corner and reverse corner , and are associated with large energy reductions .
the ionic displacements are such as to smooth out the surface irregularity .
this is easy to understand since the ( 001 ) surface of mgo crystal is the most stable one @xcite .
our displacements agree qualitatively with the displacements obtained in the model - based calculations @xcite , but in the absence of published quantitative results from the latter work we are unable to make detailed comparisons .
the most important new information from this part of our work concerns the surface electronic states .
we have shown that there is a substantial narrowing of the band - gap at the flat ( 001 ) surface by ca .
1.6 ev , and that this is entirely due to the formation of surface states at the bottom of the conduction band .
we find no surface states above the top of the valence band , and nor do we find any surface states associated with the o(2_s _ ) band .
these findings are in accord with recent lmto lda work @xcite where detailed calculations of the band structure of the ( 001 ) mgo surface have been done .
surface states pulled down from the cbm have been found there , the bottom of this band lying ca .
0.6 ev below the bottom of the bulk conduction band .
our results on the predominant localization on surface oxygens of the surface states associated with the bottom of the conduction band support recent theoretical semiempirical calculations @xcite .
namely , it was shown that surface excitonic transitions in mgo are of the one - center - type , i.e. are mainly localised on surface oxygen atoms . according to our calculations ,
the dos of the perfect step is essentially the same as for the flat surface , but important new features appear for the corner - reverse corner system .
first , a prominent band of surface states splits off from the o(2_s _ ) band . the existence of this type of surface state is well known from _ ab initio
_ calculations on other oxides such as tio@xmath52 @xcite . in the present case
, it seems to exist only at corners , and this suggests that it could be used as an experimental tool for detecting the presence of corners .
the second new feature is the appearance of surface states above the bulk vbm , which leads to a further narrowing of the surface band - gap by ca . 0.8 ev below its value for the flat surface .
this might also be used as an experimental tool for detecting corners .
the systematic reduction of surface band - gap as one passes from the bulk to the flat ( 001 ) surface and then to the corner irregularity may already have been observed in the uv diffuse reflectance spectroscopy @xcite , which indicates a series of peaks below the fundamental absorption edge ( which was found to be at 8.7 ev ) .
first two peaks at 4.62 ev and 5.75 ev have been attributed to bound ( localised ) excitonic transitions near 3-fold ( corners ) and 4-fold ( steps ) surface oxygens .
another two peaks at 6.6 ev and 7.7 ev have been associated with free excitons on the flat surface and in the bulk , respectively .
these experimental observations are generally consistent with what we have obtained in our calculations on the flat surface and various surface irregularities .
the same trend for the band gap was also found in dv - x@xmath53 cluster calculations @xcite .
turning now to our results for the f - center , we note first that our calculated properties of the bulk f - center are in close agreement with the previous calculations of wang and holzwarth @xcite .
our finding that the bulk f - center level lies 2.7 ev above the vbm is very similar to their energy of 2.3 ev , and our conclusions about the strong localization of the total electron density at the f - center site but the much weaker localization of the density associated with the f - center level are in complete agreement with what they find .
we also agree about the almost total absence of relaxation effects .
the new information we have added is that the results are completely independent of the size of repeating cell , provided the number of sites is at least 16 .
we can therefore regard our results as referring to an isolated f - center in an infinite crystal .
our calculated value of 10.5 ev for the bulk oxygen removal energy can be related to the f - center formation energy q defined as the oxygen removal energy minus the cohesive energy per formula unit with respect to free atoms @xcite . using our calculated cohesive energy of 11.1 ev
, we obtain the value @xmath54 ev .
the calculations of wang and holzwarth @xcite gave the value @xmath55 ev . in principle
, @xmath56 can be measured by additive coloring experiments in which the equilibrium between bulk f - centers and mg vapour is studied as a function of temperature and pressure .
there has been only one attempt @xcite to measure @xmath56 for mgo , which gave the value 1.5 ev .
the calculated values obtained both by us and by wang and holzwarth are clearly in poor agreement with this .
however , substantial errors are quoted in the experimental measurements , and we believe that the measured @xmath56 is subject to an error of at least @xmath57 ev . in addition , there may well have been systematic errors due to lack of equilibration between bulk and vapor .
we therefore believe that this disagreement is not necessarily due to inaccuracy in the calculations .
our calculations have shown the expected reduction of oxygen removal energy as the coordination number of the oxygen site decreases through the sequence : bulk , flat surface , reverse corner , step , corner .
the removal energy actually decreases rather little , being only 2.5 ev or 20 % lower at the corner than it is in the bulk .
other properties of the f - center also seem to depend rather weakly on its location . as we go from the bulk to the flat surface , the degree of localization of the f - center electron density ( as measured by the value of the density at the f - center site ) scarcely changes , the energy of the f - center level decreases by only 0.4 ev , and relaxation effects continue to be negligible .
the further changes in electron distribution and the energy of the f - center level on going to the step site are almost negligible .
it is only on going to the corner and reverse corner sites that significant changes occur .
namely , the f - center level is found surprisingly close to the top of the valence band for the 5-fold coordinated reverse corner system ( 0.8 ev ) . however , as follows from our calculations , for the 3-fold coordinated corner system , for which the f - center electronic density is substantially delocalised , its level is further pulled up by 0.9 ev .
the work of lnk and jmh is funded by epsrc grants gr / j37546 and gr / h67935 . the work of lnk was also supported in part by the latvian scientific council , grant 93.270 .
the calculations were performed using an allocation of time on the cray t3d at epcc provided by the high performance computing initiative . for assistance in enhancing the speed of the cetep code
we are indebted to dr .
i. bush of daresbury laboratory .
analysis of the results was performed using local distributed hardware funded by epsrc grants gr / h31783 and gr / j36266 . we gratefully acknowledge useful discussions with prof .
r. joyner , prof .
a. m. stoneham frs , dr . a. h. harker , dr .
a. l. shluger , prof .
m.spaeth and dr .
r. williams .
v. e. henrich and p. a. cox , _ the surface science of metal oxides _ ( cambridge univ .
press , 1994 ) .
v.e.henrich , rep . prog .
* 48 * ( 1985 ) 1481 .
m. tsukada , h. adachi , and c. satoko , progr . in surf .
* 14 * ( 1983 ) 113 .
e. garrone , a.zecchina , and f. s. stone , phil . mag .
b * 42 * ( 1980 ) 683 .
s. ferrer , j. m. rojo , m. salmeron , and g. a. somorjai , phil . mag . * a 45 * ( 1982 ) 261 .
s. coluccia , a. barton , and a. j. tench , j. chem .
faraday trans .
i * 77 * ( 1981 ) 2203 .
w. c. mackrodt , j. chem .
faraday trans .
ii * 85 * ( 1985 ) 541 .
e. a. colbourn , j. kendrick , and w. c. mackrodt , surf .
* 126 * ( 1983 ) 550 .
p. w. tasker and d. m. duffy , surf .
* 137 * , 91 ( 1984 ) .
v. puchin , a. shluger , y. nakai , and n. itoh , phys .
b * 49 * ( 1994 ) 11364 .
j. goniakowski and c. noguera , surf .
, submitted .
l. shluger , j. d. gale , and c. r. a. catlow , j. phys .
* 96 * ( 1992 ) 10389 .
w. langel and m. parrinello , phys .
* 73 * ( 1994 ) 504 .
u.schnberger and f.aryasetiawan , phys .
b , submitted . s. pugh and m. j. gillan , surf .
* 320 * ( 1994 ) 331 .
m. caus , r. dovesi , c. pisani , and c. roetti , surf .
* 175 * ( 1986 ) 551 .
m. caus , r. dovesi , e. kotomin , and c. pisani , j. phys .
c : solid state phys . * 20 * ( 1987 ) 4983 .
a. gibson , r. haydock , and j. p. lafemina , j. vac .
technol . * a 10 * ( 1992 ) 2361 . c. a. scamehorn , n. m. harrison , and m. i. mccarthy , j. chem .
* 101 * ( 1994 ) 1547 .
g. p. srivastava and d. weaire , adv .
* 36 * ( 1987 ) 463 .
m. j. gillan , in proc .
nato asi on computer simulation in material science , aussois , march 1991 , ed .
m. meyer and v. pontikis , p. 257
( dordrecht , kluwer , 1991 ) m. c. payne , m. p. teter , d. c. allan , t. a. arias , and j. d. joannopoulos , rev . mod .
* 64 * ( 1992 ) 1045 .
i. manassidis and m. j. gillan , j. am .
* 77 * ( 1994 ) 335 .
i. manassidis , a. de vita , and m. j. gillan , surf .
* 285 * ( 1993 ) l517 .
a. de vita , i. manassidis , j .- s .
lin , and m. j. gillan , europhys .
* 19 * ( 1992 ) 605 .
a. de vita , m. j. gillan , j. s. lin , m. c. payne , i. tich , and l.j.clarke , phys . rev .
b * 46 * ( 1992 ) 12964 . k. j. chang and m. l. cohen , phys .
b * 30 * ( 1984 ) 4774 .
q. s. wang and n. a. w. holzwarth , phys .
b * 41 * ( 1990 ) 3211 . b. m. klein , w. e. pickett , l. l. boyer , and r. zeller , phys . rev .
b * 35 * ( 1987 ) 5802 .
k. jackson , m. r. pederson , and b. m. klein , phys .
b * 43 * ( 1991 ) 2364 . a. gibson , r. haydock , and j. lafemina , phys .
b * 50 * ( 1994 ) 2582 . yong - nian xu and w. y. ching , phys .
b * 43 * ( 1991 ) 4461 .
r. car and m. parrinello , phys .
lett . * 55 * ( 1985 ) 2471 . l. j. clarke , i. tich , and m. c. payne , comp .
* 72 * ( 1992 ) 14 .
l. kleinman and d. m. bylander , phys .
* 48 * ( 1980 ) 566 .
r. d. king - smith , m. c. payne , and j. s. lin , phys .
b * 44 * ( 1991 ) 13063 .
d. m. ceperley and b. j. alder , phys .
* 45 * ( 1980 ) 566 .
j. perdew and a. zunger , phys . rev .
b * 23 * ( 1981 ) 5048 .
h. j. monkhorst and j. d. pack , phys .
rev b * 13 * ( 1976 ) 5188 .
r. w. g. wyckoff , _ crystal structures _ , 2nd ed .
2 ( interscience , new york , 1964 ) ch .
g. lehmann and m. taut , phys .
. sol . * 54 * ( 1972 ) 469 .
o. jepsen and o. k. andersen , solid state comm .
* 9 * ( 1971 ) 1763 .
p. e. blchl , o. jepsen , and o. k. andersen , phys .
b * 49 * ( 1994 ) 16223 . c. e. moore , _ atomic energy levels _ , nat . bur .
( u.s . ) ref .
gpo , washington , d.c . , 1971 ) , vol .
o. e. taurian , m. springborg , and n. e. christensen , solid state comm . * 55 * ( 1985 ) 351 . ,
60th ed . , edited by r. c. weast and m. j. astle ( chemical rubber co. , boca raton , fl , 1980 ) , p. d-72 . j. vail , j. phys .
solids * 51 * ( 1990 ) 589 .
r. pandey and j. vail , j. phys .
matter * 1 * ( 1989 ) 2801 .
a. b. sobolev , v. a. lobach , and b. v. shulgin , sov .
phys . solid state * 27 * ( 1986 ) 1881 .
r. pandey , j. e. jaffe , and a. b. kunz , phys .
b * 43 * ( 1991 ) 9228 .
s. p. kowalczyk , f. r. mcfeely , l. ley , v. t. gritsyna , and d. a. shirley , solid state comm .
* 23 * ( 1977 ) 161 .
l. fiermans , r. hoogewijs , g. de mayer , and j. vennik , phys .
status solidi a * 59 * ( 1980 ) 569 .
m. l. bortz , r. g. french , d. j. jones , r. v. kasowski , and f. s. ohuchi , phys .
* 41 * ( 1990 ) 537 .
d. w. fisher , adv .
x - ray analysis * 13 * ( 1970 ) 159 .
d. m. roessler and w. c. walker , phys .
* 159 * ( 1967 ) 733 .
l. a. kappers , r. l. kroes , and e. b. hensley , phys .
b * 1 * ( 1970 ) 4151 .
y. chen , v. m. orera , r. gonzalez , r. t. williams , g. p. williams , g. h. rozenblatt , and g. j. pogatshnik , phys .
b * 42 * ( 1990 ) 1410 .
y. chen , j. l. kolopys , and w. a. sibley , phys . rev .
* 186 * ( 1969 ) 865 .
g. p. summers , t. m. wilson , b. t. jeffries , h. t. tohver , y. chen , and m. m. abraham , phys . rev .
b * 27 * ( 1983 ) 1283 .
g. h. rosenblatt , m. w. rowe , g. p. williams , jr . , and
r. t. williams , phys . rev .
b * 39 * ( 1989 ) 10309 .
t. urano , t. kanaji , and m. kaburagi , surf .
* 134 * ( 1983 ) 109 .
d. l. blanchard , d. l. lessor , j. p. lafemina , d. r. baer , w. k. ford , and t. guo , j. vac .
technol . a * 9 * ( 1991 ) 1814 .
w. r. welton - cook and w. berndt , j. phys .
c : solid state phys . * 15 * ( 1982 ) 5691 .
a. j. martin and h. bilz , phys .
b * 19 * ( 1979 ) 6593 .
j. goniakowski and c. noguera , surf .
* 323 * ( 1995 ) 129 .
v. e. henrich , g. dresselhaus , and h. j. zeiger , phys .
b * 22 * ( 1980 ) 4764 .
a. gibson , r. haydock , and j. p. lafemina , appl . surf .
* 72 * ( 1993 ) 285 .
d. vogtenhuber , r. podloucky , a. neckel , s. g. steinemann , and a. j. freeman , phys .
b * 49 * ( 1994 )
2099 . j. magill , j. bloem , and r. w. ohse , j. chem . phys .
* 76 * ( 1982 ) 6227 .
v. t. coon , surf .
* 88 * ( 1979 ) l42 .
v. m. bermudez , surf .
* 74 * ( 1978 ) 568 .
a. l. shluger , r. w. grimes , c. r. a. catlow , and n. itoh , j. phys . : condens .
matter * 3 * ( 1991 ) 8027 .
2 . a contour plot of the valence electronic density of the 54-site f - center system ( in units of @xmath60 electron / @xmath24 ) .
the cut has been made in the ( 001 ) plane . to avoid high peaks on oxygens ,
the density has been chopped at 0.2 electron / @xmath24 .
the positions of oxygen atoms are indicated by the symbol x. distances are in .
4 . a contour plot of the electronic density associated with the f - center band for the 54-site f - center system .
positions of mg atoms are indicated by the symbol @xmath61 .
the units and other notations are the same as in fig .
2 . fig .
5 . the total dos of the unrelaxed six - layer slab ( top panel ) and the flat surface systems containing third layer ( middle panel ) and first layer ( bottom panel ) f - centers .
arbitrary units are used for the dos . fig . 6 . a contour plot of the fictitious charge density ( see eq .
( 3 ) ) associated with the first unoccupied state at the bottom of the conduction band of the flat surface .
the cut has been made parallel to the surface normal through nearest surface mg and o ions .
the top two layers of atoms are shown explicitly .
the units and notations are as in fig .
4 . fig . 7 .
the contour plot of the valence electronic density of the flat surface system containing the first - layer f - center .
the cut has been made along the direction of the surface normal through the f - center and nearest mg ions on the surface .
other notations and units are the same as in fig .
2 . fig .
8 . repeating geometries used in the _ ab initio _ calculations for the step ( _ a _ ) and the corner - reverse corner ( _ b _ ) systems .
mg and o atoms are indicated by grey and white circles , respectively . in each case , only upper side of the slab is shown .
note that the axes of the cartesian coordinate system are also shown for each case .
they correspond to eqs.(4 ) and ( 5 ) .
corresponding f - center systems have been created by removing oxygen atoms 2 ( _ a _ ) , 3 or 1 ( _ b _ ) in the case of the step , corner and reverse corner systems , respectively .
9 . the electronic dos of the perfect corner - reverse corner system ( top panel ) , the system containing the f - center at the corner ( middle panel ) and the reverse corner ( bottom panel ) .
arbitrary units are used for the dos .
fig . 10 . a contour plot of the fictitious charge density ( see eq .
( 3 ) ) associated with the band of unoccupied states split off from the the bottom of the conduction band of the perfect corner - reverse corner system . the cut has been made in a plane parallel to the terraces passing through the corners along the zig - zag step .
other notations and units are the same as in fig .
4 . fig . 13 . a contour plot of the electronic density associated with the f - center band for the corner f - center system .
the cut is the same as in fig .
10 . other notations and units
are the same as in fig . | _ ab initio _ calculations based on the density - functional pseudopotential approach have been used to study the fully relaxed structure , the electron distribution and the electronic density of states of ( 001 ) terraces , steps , corners and reverse corners , and of f - centers at these surface features on mgo .
the calculations confirm earlier predictions of the relaxed structures of surface irregularities based on simple interaction models
. a substantial narrowing of the band - gap is found at the surface , which for terraces and steps is due to surface states at the bottom of the conduction band , but for the corner and reverse corner is also due to surface states at the top of the valence band .
the f - center formation energy decreases steadily as the coordination of the oxygen site is reduced .
the energy of the f - center level shows a tendency to approach the top of the valence band as the coordination of its site decreases .
22 cm 15.5 cm 1.0 cm 1.0 cm -1.0 cm = 5000 = 10000 = 5000 = 2000 = 5000 = 0.5pt |
the paper introduces a product construction of legendrian submanifolds in contact manifolds of the form @xmath0 .
it gives a multitude of examples of interesting legendrian submanifolds in a wide array of contact manifolds , not simply affine space and jet spaces , and generalizes some previous constructions .
this construction also seems useful for technical reasons and allows the construction of relative invariants of legendrian knots and further study of spaces of legendrian knots .
we also compute the thurston - bennequin invariant and maslov class for these product legendrians in terms of geometric and topological properties of the factors and obtain an explicit formula when the ambient contact manifold is @xmath3 .
this calculation demonstrates that the legendrian isotopy class of the product is determined by the particular embedding of a legendrian factor and not its isotopy class .
it shows that the contact geometry of these spaces is extremely rich . the motivation for this paper is exploring the contact geometry of manifolds @xmath0 , where @xmath4 is an exact symplectic manifold , by constructing many examples of legendrian submanifolds .
the legendrian submanifolds of @xmath5 , legendrian knots , are well - explored and have found importance in low - dimensional topology .
however , much less is known about higher dimensional ( @xmath6 ) contact manifolds and their legendrian submanifolds .
several authors have built examples , including the frontspinning construction of ekholm , etnyre and sullivan @xcite , a generalization to spheres of arbitrary dimensions by golovko @xcite , hypercube tori defined by baldridge and mccarty @xcite and tori given as the trace of legendrian isotopies by ekholm and kalman @xcite .
in fact , the former three construction should naturally be understood as products in the sense of this paper . as smooth knotting is a codimension 2 phenomenom , legendrian submanifolds in higher dimensions
lose some of their relevance to smooth topology of their ambient spaces .
yet there are indications that the contact geometry of these spaces is rich with legendrian submanifolds and this paper gives more evidence for this suspicion .
for instance , knot contact homology is an invariant of smooth knots in @xmath5 , defined by lenhard ng , that assigns to each knot a legendrian torus in @xmath7 .
legendrian contact homology is a legendrian invariant of the torus and is therefore a smooth invariant of the underlying knot . as of this writing , there are currently no known examples of two inequivalent smooth knots with isomorphic knot contact homology .
the construction of legendrian products is straightforward . for a detailed explanation of the terminology , see section [ sec : prelim ] .
let @xmath8 and @xmath9 be contact manifolds such that @xmath10 are exact symplectic manifolds and with contact forms @xmath11 , @xmath12 , respectivly .
take legendrian submanifolds @xmath13 and @xmath14 with reeb chords @xmath15 and let @xmath16 denote their lagrangian projections in @xmath17
. then @xmath18 is an exact lagrangian submanifold of @xmath19 .
[ def : legprod ] the _ legendrian product _ @xmath2 is the legendrian submanifold in @xmath20 given by the lift of @xmath18 .
it is an immersed legendrian submanifold and is embedded if the sets of reeb chord actions @xmath21 are disjoint .
* caution*. this construction is _ not _ well - defined up to legendrian isotopy of one of the factors . if @xmath22 is a continuous , one - paramter family of embedded legendrians in @xmath0 , it is not true that @xmath23 gives a legendrian isotopy through embeddings .
legendrian isotopies can introduce or annihilate reeb chords and change reeb chord actions , which may introduce self - intersections of @xmath23 for some @xmath24 .
thus the product construction depends upon the explicit embeddings @xmath25 and @xmath26 and not simply on the legendrian isotopy class of @xmath27 or @xmath28 . in fact , by varying the legendrian embedding of a factor within its legendrian isotopy class ,
one can obtain infinitely many , non - legendrian isotopic products .
[ thrm : infty ] let @xmath29 be chord generic legendrians such that @xmath30 have different parity
. then there exists an infinite family of legendrians @xmath31 all legendrian isotopic to @xmath27 such that the family of legendrian products @xmath32 are pairwise non - legendrian isotopic . to prove this
, we calculate the thurston - bennequin invariant of these legendrian products in terms of the embeddings of @xmath1 .
there are two well - known , classical invariants of legendrian knots : the thurston - bennequin number and the rotation number .
these have been generalized to higher dimensions by tabachnikov and by ekholm , etnyre and sullivan .
the latter described the thurston - bennequin invariant for homologically - trivial legendrians @xmath28 as the linking number of @xmath28 with a pushoff @xmath33 of itself along the reeb vector field @xmath34 in our setting , when the ambient contact manifold is @xmath0 , this can be computed in a manner similar to the writhe of knots .
the maslov class is a cohomology class @xmath35 that assigns to each 1-dimensional homology class the maslov index of a path representing that class . in dimension 3 ,
the rotation number is @xmath36 where @xmath37 is a generator of @xmath38 .
suppose that @xmath39 and @xmath40 are chord generic legendrians with reeb chords @xmath15 .
then we can obtain the following formula for the classical invariants of their product .
[ thrm : tb ] the thurston - bennequin number of @xmath2 is given by : @xmath41 where @xmath42 the maslov class of @xmath2 is given by @xmath43 the last term in the @xmath44 formula requires some elaboration .
locally , each transverse double point is the intersection of two open @xmath45-disks in a single point .
just as computing the sign of a knot crossing requires knowing which strand passes over the other , computing the sign of a reeb chord requires knowing which disk passes `` over '' the other . in the product @xmath2 , after a suitable perturbation , there is a reeb chord @xmath46 for each pair @xmath47 of reeb chords of @xmath27 and @xmath28 .
however , determining which disk is `` over '' and which is `` under '' depends upon the relative lengths of the reeb chords @xmath48 . using this product construction , it is also possible to build examples of legendrian submanifolds in closed contact manifolds .
let @xmath49 be a contact manifold , @xmath50 a legendrian submanifold , @xmath51 a contact embedding of codimension @xmath52 with trivial conformal symplectic normal bundle and @xmath53 a legendrian submanifold .
then we can identify a neighborhood of @xmath28 with a neighborhood of the 0-section in @xmath54 and find @xmath55 as a legendrian submanifold of @xmath56 , which we denote by @xmath57 and refer to as the _ spinning _ of @xmath27 by @xmath28 .
it is clear from the construction that this is invariant under legendrian isotopies of @xmath28 in @xmath49 , contact isotopies of @xmath49 in @xmath56 and compactly supported isotopies of @xmath27 . in @xcite , etnyre , ekholm and sullivan
defined a construction , called frontspinning , which takes a legendrian @xmath58 and produces a legendrian @xmath59 of topological type @xmath60 .
in fact , these should more naturally be seen as the product of @xmath28 with a standard legendrian unknot with @xmath61 .
consider @xmath62 as the unit sphere in @xmath63 and take the obvious contact embedding @xmath64 .
we prove the following theorem and corollary .
[ thrm : spin ] let @xmath28 be a legendrian submanifold of @xmath62 ( equiv @xmath65 ) and @xmath27 a legendrian submanifold of @xmath66 , chosen so that all reeb chord actions of @xmath27 are much less than all reeb chord actions of @xmath28
. then @xmath67 and @xmath2 are legendrian isotopic in @xmath3 let @xmath68 denote the whitney embedding of the standard legendrian unknot of @xmath69 with exactly one reeb chord of length 1 .
[ cor : front ] let @xmath58 be legendrian all of whose reeb chords have action @xmath70 . then @xmath71 and @xmath72 are legendrian isotopic in @xmath3 .
golovko has recently @xcite extended the frontspinning construction to produce legendrians in @xmath73 of topological type @xmath74 and this can be interpreted as products with whitney spheres of arbitrary dimension .
the thurston - bennequin invariant is sometimes blind , as in @xcite it is shown to reduce to the euler characteristic when the ambient manifold is @xmath3 for even @xmath45 .
so the thurston - bennequin invariant is not useful for distinguishing products when the dimensions of the factors have the same parity .
the results here suggest that a more powerful invariant , such as legendrian contact homology , will reveal many interesting examples of legendrian submanifolds in higher dimensional contact manifolds .
legendrian contact homology is an invariant of legendrian submanifolds that fits into the general framework of symplectic field theory .
it has been constructed by ekholm , etnyre and sullivan when the ambient contact manifold is some @xmath0 @xcite , @xcite . to each legendrian @xmath28 , it associates a diferential graded algebra , which is invariant up to an appropriate algebraic equivalence .
the algebra is the unital tensor algebra over the group ring @xmath75 $ ] generated by the reeb chords of @xmath28 .
the differential counts rigid , punctured holomorphic maps @xmath76 with a single positive puncture and arbitrarily many negative punctures . in dimension 3
, the legendrian contact homology of legendrian knots is known as the chekanov - eliashberg dga and by the riemann mapping theorem , can be computed combinatorially from either the front or lagrangian projection of a knot .
the legendrian contact homology of the product @xmath55 can be computed in terms of geometric and topological data of the factors ( see @xcite ) .
holomorphic disks on the product @xmath55 are determined by and can be constructed from holomorphic disks and gradient flow lines on the factors .
however , an interesting complication arises from the dependence on relative reeb chord lengths that first appeared in the thurston - bennequin calculation .
specifically , the lch of the product potentially includes information determined by holomorphic disks with multiple positive punctures .
this information is ignored in the standard version of legendrian contact homology .
this suggests that a full understanding of the lch of products is more interesting than one might expect from the construction and from results of ekholm , etnyre and sabloff on the linearized contact homology of frontspun legendrians @xmath71 @xcite .
given that the lch of legendrian knots can be computed combinatorially and that ekholm has established a correspondence between rigid holomorphic disks on @xmath77 and rigid gradient flow trees on @xmath28 , it is reasonable to conjecture the following : the legendrian contact homology of the products of legendrian knots can be computed combinatorially .
finally , for @xmath78 and @xmath53 suitably small , theorem [ thrm : spin ] shows that the product @xmath55 is well - defined up to legendrian isotopy of the factors .
thus , the legendrian contact homology of the product @xmath55 is a legendrian invariant of both of the factors .
yet since it depends upon geometric data not included in the lch of the factors , this invariant may provide more information about the legendrian isotopy class of @xmath28 and @xmath27 than the lch of the factors individually .
( relative legendrian invariants ) is lch(@xmath55 ) a stronger invariant of @xmath28 than @xmath79 for some @xmath27 ?
in particular , one may fix @xmath28 and vary @xmath27 , using the lch of the product to probe the contact geometry of @xmath28 .
the outline of this paper is as follows .
section [ sec : prelim ] reviews some preliminary content and definitions about contact geometry , legendrians , reeb chords in the relevant setting of manifolds of the form @xmath0 .
section [ sec : tb ] contains the proofs of theorems [ thrm : tb ] and [ thrm : infty ] .
this first requires constructing a suitable perturbation of the conormal lift as it is not chord generic .
section [ sec : spin ] defines spinning and proves theorem [ thrm : spin ] and [ cor : front ] .
section [ sec : exam ] discusses examples , including products of whitney spheres and products of legendrian knots .
i would like to thank my advisor , scott baldridge , for advice and encouragement , as well as john etnrye , forrest gordon , kate kearney , ben mccarty and shea vela - vick for helpful conversations .
let @xmath4 be an exact symplectic manifold of dimension @xmath80 , meaning the symplectic form @xmath81 can be expressed as @xmath82 some primitive @xmath83 .
a submanifold @xmath84 is _ isotropic _ if the restriction @xmath85 is identically 0 and _ lagrangian _ if it is isotropic and has dimension @xmath45 .
a lagrangian submanifold is _ exact _ if the restriction @xmath86 is exact . for any ( immersed )
lagrangian @xmath28 , there is an oriented diffeomorphism between the normal bundle @xmath87 of @xmath28 in @xmath4 and the cotangent bundle @xmath88 .
moreover , there is a symplectomorphism between some neighborhood of @xmath28 in @xmath4 and a neighborhood of the 0-section in @xmath88 .
the 1-form @xmath11 is a contact form on the product manifold @xmath0 ( @xmath89 is the coordinate on the second factor ) as @xmath90 .
this contact form induces a contact structure @xmath91 , which is a maximally nonintegrable hyperplane field on @xmath0 , and a _ reeb vector field _ @xmath92 .
the contact manifold @xmath93 is called the _ contactization _ of @xmath4 .
a submanifold @xmath28 of some contact manifold @xmath94 is _ isotropic _ if it is everywhere tangent to the hyperplane field @xmath95 and is _ legendrian _ if it is isotropic and has dimension @xmath45 .
a continuous , one - parameter family of legendrian submanifolds is a _
legendrian isotopy_. let @xmath96 be the projection map onto the first factor , which is called the _
lagrangian projection_. throughout this paper , we distinguish points and sets @xmath97 from their images under @xmath98 through bar notation i.e @xmath99 .
the projection @xmath100 of a legendrian submanifold is exact lagrangian , since for @xmath101 , the restriction of the projection @xmath102 to @xmath28 , the legendrian condition implies that @xmath103 .
furthermore , given an exact lagrangian submanifold @xmath100 in @xmath4 , there exists a _ lift _
@xmath104 of @xmath100 such that @xmath28 is an immersed legendrian submanifold .
this can be chosen uniquely up to a translation in the @xmath89-direction .
a _ reeb chord _
@xmath105 for some legendrian submanifold @xmath28 is an integral curve of the reeb vector field that begins and ends on @xmath28 .
since the reeb vector field flows in the @xmath89-direction , the endpoints @xmath106 of the chord project to the same point @xmath107 and every multiple point of the projection @xmath108 lifts to at least one reeb chord .
a legendrian submanifold is _ chord generic _ if its lagrangian projection has a finite number of transverse double points .
double points of @xmath108 correspond to unique reeb chords since @xmath109 is not compact and this implies that @xmath28 has a finite number of reeb chords .
the _ action _ @xmath110 of a reeb chord is its length , which is equal to the difference of @xmath89-coordinates @xmath111 .
choose neighborhoods @xmath112 around @xmath106 and call these the _ upper sheet _ and _ lower sheet_. each reeb chord has a sign @xmath113 defined as follows .
let @xmath114 and @xmath115 . since @xmath105 corresponds to a transverse double point in the lagrangian projection , @xmath116 span @xmath117 . if the orientation of @xmath116 agrees with the orientation of @xmath4 , then @xmath118 ; otherwise @xmath119 .
let @xmath120 be an oriented contact manifold with contact structure @xmath121 and recall that the 2-form @xmath122 restricts to a symplectic form on @xmath95 . if a submanifold @xmath28 is isotropic , then @xmath123 and define @xmath124 to be the symplectic subbundle of @xmath125 whose fibers are the symplectic orthogonal complements to the fibers of @xmath126 .
the _ conformal symplectic normal bundle _ of @xmath28 in @xmath49 is the quotient bundle @xmath127 if dim@xmath128 and dim @xmath129 then @xmath130 has rank @xmath131 .
if we choose an almost complex structure @xmath132 on @xmath95 compatible with @xmath122 , then the normal bundle of @xmath28 in @xmath49 splits as @xmath133 the contact form restricts to a contact form on the fibers of @xmath134 and so it is a contact subbundle of @xmath135 .
furthermore , there exists a contactomorphism between suitable neighborhoods of @xmath50 and the 0-section of @xmath136 .
similarly , given a contact embedding @xmath137 , the _ conformal symplectic normal _
bundle @xmath138 is the symplectic subbundle @xmath139 given by taking the symplectic orthogonal complement to @xmath95 in @xmath140 . this bundle can be identified with the normal bundle @xmath141 of @xmath49 in @xmath56 .
there also exists a contactomorphism between suitable neighborhoods of @xmath142 and the 0-section of @xmath138 .
let @xmath143 be the grassman manifold of lagrangian subspaces in the standard affine symplectic space @xmath144 .
fix some lagrangian subspace @xmath145 and let @xmath146 be the set of all lagrangian planes in @xmath147 that intersect @xmath148 along a subspace of dimension @xmath149 .
then the _ maslov cycle _ is the algebraic subvariety @xmath150 which has codimension 1 in @xmath143 .
for a path @xmath151 \rightarrow \lambda_n$ ] , we can define an intersection number of @xmath152 and @xmath153 as follows .
fix a lagrangian complement @xmath68 to @xmath148 and suppose that @xmath154 intersects @xmath153 . for @xmath24 near @xmath155 ,
there exists a family @xmath156 of vectors such that for all @xmath157 the vector @xmath158 .
then there is a quadratic form @xmath159 on @xmath160 and the signature of this quadratic form is the intersection number of @xmath152 at @xmath155 . if @xmath152 is a loop , then the _ maslov index _
@xmath161 is the total intersection number of @xmath152 with @xmath153 .
the map @xmath162 defines an isomorphism @xmath163 .
let @xmath164 be an immersed lagrangian .
a global trivialization of @xmath165 induces a map @xmath166 , where each point is sent to its lagrangian tangent plane .
maslov class _ is the pullback of a generator @xmath167 of @xmath168 @xmath169
the goal of this section is to prove theorems [ thrm : tb ] and [ thrm : infty ] .
one problem with this product is that while each factor is chord generic , the product is not .
in fact , each reeb chord is part of some family of reeb chords given by either @xmath170 or @xmath171 . in order to make @xmath2 chord generic
, we must perturb it slightly .
let @xmath172 be @xmath173-small morse functions on @xmath1 whose critical points are away from the endpoints of the reeb chords and such that the endpoints lie in different level sets ( i.e. @xmath174 ) .
thus , there exist neighborhoods @xmath175 of @xmath176 such that @xmath177 and @xmath178 are disjoint and similarly there exist such neighborhoods @xmath179 of each @xmath180 .
denote the critical points of @xmath181 by @xmath182 and the critical points of @xmath183 by @xmath184 .
we can identify a small neighborhood of @xmath2 with a neighborhood of the 0-section in its 1-jet space @xmath185 and perturb it by a legendrian isotopy to the graph of @xmath186 in the 1-jet space @xmath185 .
the perturbed legendrian is chord generic and has the following reeb chords : * * reeb / morse * : one for each pair @xmath187 of reeb chord for @xmath27 , morse critical point of @xmath28 , denoted @xmath188 * * morse / reeb * : one for each pair @xmath189 of morse critical point of @xmath27 , reeb chord for @xmath28 , denoted @xmath190 * * reeb / reeb * : two for each pair @xmath191 of reeb chord for @xmath27 , reeb chord for @xmath28 , denoted @xmath46 and @xmath192 we will refer to these as a - chords , b - chords , c - chords and d - chords , respectively . in the lagrangian projection ,
a neighborhood of @xmath193 is symplectomorphic to a neighborhood @xmath194 of the 0-section of the cotangent bundle .
moreover , we can assume that the map @xmath195 is injective away from @xmath196 .
the perturbation pushes @xmath2 off to the graph of @xmath197 .
let @xmath198 denote points in @xmath1 and @xmath199 denote those points projected to @xmath17 .
then the perturbation maps @xmath200 in @xmath19 to @xmath201 .
now , suppose that @xmath202 for some @xmath203 .
if @xmath204 , then either @xmath205 is a morse critical point or @xmath206 lie in the same level set . in the first case
, we get intersection points coming from the intersection points of @xmath207 , the pushoff , whose intersection points are in 1 - 1 correspondance with the intersection points of @xmath28 .
this gives the a - chords .
the second is impossible , since if @xmath206 are distinct then we must have ( up to relabeling ) that @xmath208 and @xmath209 and so @xmath206 can not lie in the same level set . repeating this for @xmath210 will yield the b - chords .
finally , consider when both pairs @xmath211 and @xmath206 are distinct .
prior to the perturbation , there was a unique transverse intersection point of @xmath212 and @xmath213 .
similarly , there was a unique transverse intersection point of @xmath214 and @xmath215 .
since the perturbation is @xmath173-small , we can assume that after the perturbation , there remains a unique intersection point in each case .
thus , the intersection point @xmath216 under consideration must be one of these two ; the first case we label @xmath192 and the second we label @xmath46 .
we can compute the signs of the intersection points as well : [ lem : sign ] the reeb chords of the perturbation have the following actions * @xmath217 * @xmath218 * @xmath219 * @xmath220 and signs * @xmath221 * @xmath222 * @xmath223 * @xmath224 where @xmath225 as above , let @xmath226 and @xmath227 , which we think of as the tangent planes to the upper and lower sheets @xmath228 in @xmath4 at @xmath229 , and define @xmath230 similarly for @xmath28 .
then for the reeb / morse chords , the tangent planes of the upper and lower sheets at @xmath231 are given by @xmath232 and @xmath233 .
the sign of the reeb chord is given by the orientation of @xmath234 , which is @xmath235 times the orientation given by @xmath236 , whose sign is given by @xmath237 .
the situation is similar for the morse / reeb chords and each @xmath192 .
however , for the @xmath46 , the calculation is different . if @xmath238 , the tangent plane to the upper sheet at @xmath239 is given by @xmath240 and to the lower sheet by @xmath241 .
thus , the orientation on @xmath242 is @xmath243 times that of @xmath244 , which is given by @xmath245 .
but if @xmath246 , then the sign of the reeb chord is given by the orientation of @xmath247 , which differs from @xmath245 by @xmath248 .
we can now prove theorem [ thrm : tb ] .
suppose that @xmath108 , the projection of @xmath28 to @xmath4 , has a finite number of transverse double points .
then these double points are in one - to - one correspondance with the reeb chords and @xmath28 is called chord generic . each chord
@xmath105 can then be assigned a sign @xmath249 and in @xcite , it is shown that the thurston - bennequin invariant is given by @xmath250 proof of theorem [ thrm : tb ] the thurston - bennequin calculation follows directly from lemma [ lem : sign ] by summing over all indices @xmath251 . in order to calclate the maslov class
, we can make the necessary choices so that each condition splits . specifically , take @xmath252 and @xmath253 and associated maslov cycles
then @xmath255 is lagrangian in @xmath256 and so defines a maslov cycle @xmath257 that splits as @xmath258 moreover , we can choose lagrangian complements such that @xmath259 .
take a path @xmath260 and its projections @xmath261 . at each intersection point
@xmath262 , the signature of the associated quadratic form is the sum of the signatures of the associated quadratic forms for the intersections of @xmath263 and @xmath254 .
thus , the maslov index splits as @xmath264 .
therefore , it follows that the maslov class @xmath265 splits as well .
notice that this formula is consistent with the result in @xcite that for @xmath28 of even dimension , the thurston - bennequin number is @xmath266 , where @xmath267 is the oriented normal bundle to @xmath28 in the lagrangian projection .
suppose that the dimensions of @xmath27 and @xmath28 have the same parity .
then the dimension of their product is even and so this result applies .
if @xmath30 are odd , then the euler characteristic of both manifolds vanish by poincare duality , the euler characteristic of their product vanishes , and @xmath268 is always negative .
thus , we get that : @xmath269 if @xmath30 are even , then @xmath268 is always 1 and we can use the immersed version of the lagrangian neighborhood theorem to identify @xmath270 with @xmath271 and @xmath88 with @xmath272 and obtain @xmath273 since it is clear from the construction that the normal bundle of the product is the product of the normal bundles .
proof of theorem [ thrm : infty ] choose some darboux ball of radius @xmath274 and isotope @xmath27 so that its intersection with the darboux ball is two disjoint disks given by two parallel lagrangian planes of distance @xmath275 apart . by a hamiltonian isotopy supported in the darboux ball , we can add two canceling transverse double points corresponding to two reeb chords @xmath276 , labeled so that @xmath277 .
now , by scaling either @xmath27 or @xmath28 and a perturbation , we can assume that there is exactly one reeb chord @xmath278 of @xmath28 such that @xmath279 .
for all other reeb chords @xmath280 of @xmath28 , the terms @xmath281 and @xmath282 cancel .
however , @xmath283 and @xmath284 have the same sign since the pairs @xmath285 and @xmath286 are distinct .
moreover , we can add aribtrarily many pairs of reeb chord pairs @xmath287 so that @xmath288 and @xmath289 for all @xmath290 .
thus , we can add @xmath291 to the thurston - bennequin invariant for arbitrary nonnegative integer @xmath45 .
the goal of this section is to prove theorem [ thrm : spin ] and corollary [ cor : front ] .
we first define frontspinning as in @xcite , then give an alternate , invariant description of this construction called spinning before proving that spinning and the legendrian product are equivalent in @xmath3 .
frontspinning is a construction defined by ekholm , etnyre and sullivan that takes a legendrian @xmath78 and produces a legendrian @xmath292 of topological type @xmath60 .
suppose that a legendrian @xmath78 is given by the embedding @xmath293 and parametrized such that @xmath294 let @xmath295 be the projection onto the @xmath205 and @xmath89 coordinates .
the _ front projection _ of @xmath28 is the subsvariety @xmath296 if @xmath297 is a subvariety of @xmath298 such that @xmath299 for any @xmath300 , then @xmath297 lifts to an isotropic subvariety of @xmath3 as the @xmath301-coordinates can be chosen at each point to satisfy the contact condition .
the _ frontspinning _ of @xmath28 is the lift to @xmath302 of the following subvariety of @xmath303 : @xmath304 for @xmath305 $ ] .
this can be thought of as spinning the front projection of @xmath28 around the plane @xmath306 in @xmath303 and the intersections of @xmath297 with the planes @xmath307 give a family of front projections of @xmath28 as @xmath308 varies .
we now give an alternate interpretation of this construction that generalizes to arbitrary contact manifolds .
choose some legendrian @xmath309 , a contact embedding @xmath51 of codimension @xmath52 with trivial @xmath138 and a legendrian @xmath310 . by scaling and translation
, we can assume that @xmath28 lies in a suitably small neighborhood of the origin .
it follows that the conformal symplectic normal bundle of @xmath27 in @xmath56 splits as @xmath311 and by assumption , since @xmath27 is legendrian , this bundle is trivial of rank @xmath52 .
the _ spinning _ of @xmath28 by @xmath27 , denoted @xmath57 , is the legendrian submanifold of @xmath56 obtained as the image of @xmath2 under the contactomorphism that identifies neighborhoods of @xmath27 in @xmath312 and @xmath56 .
it is clear from the construction that ambient contact isotopies of @xmath313 and @xmath28 extend to legendrian isotopies of @xmath57 , provided that @xmath28 is contained in a suitably small neighborhood of 0 . consider @xmath314 as the unit sphere in @xmath315 .
then for the embedding @xmath316 , the conformal symplectic normal bundle @xmath317 is trivial .
let @xmath68 be the submanifold @xmath318 it is easy to verify that w is legendrian and isotopic to the standard legendrian unknot with @xmath61 .
[ lemma : sigmal ] the legendrian submanifolds @xmath71 and @xmath319 are identical . let @xmath132 be the standard complex structure on @xmath320 .
then we can trivialize @xmath321 as @xmath322 @xmath323 and trivialize @xmath324 as @xmath325 the reeb vector field in @xmath69 along @xmath68 is given by @xmath326 .
it follows from above that the frontspun @xmath71 is obtained by restricting to @xmath28 in each fiber of the bundle @xmath327 we can now prove theorem [ thrm : spin ] choose a darboux ball around the point @xmath328 , given by some map @xmath329 of the unit ball , that restricts to a darboux ball @xmath330 on @xmath62 as well .
thus , @xmath331 is exactly @xmath332 .
choose some legendrian isotopy @xmath22 in @xmath62 so that @xmath333 and @xmath334 in this darboux ball on @xmath62 .
the isotopy extends to some suitable neighborhood of @xmath27 and isotope @xmath28 so that @xmath67 lies in this neighborhood .
the legendrian isotopy of @xmath27 thus extends to a legendrian isotopy from @xmath67 to @xmath331 .
recall that the contact disk theorem states that for any two contact embeddings @xmath335 of the unit ball into a contact manifold , there is a contact isotopy @xmath336 such that @xmath337 .
thus , there exists some isotopy @xmath338 such that @xmath339 and this isotopy sends @xmath331 to @xmath2 .
furthermore , since @xmath28 must live in a suitably small neighborhood of the origin , whose diameter must be less than the length of the reeb chords of @xmath27 , it follows that all its reeb chord actions must be less than this diameter .
corollary [ cor : front ] now follows directly from lemma [ lemma : sigmal ] and theorem [ thrm : spin ] .
a whitney sphere @xmath340 is the legendrian sphere @xmath341 given as the lift of the following exact lagrangian immersion in @xmath342 : @xmath343 where @xmath105 is some positive real constant .
this immersion has exactly 1 transverse double point at @xmath344 and so the legendrian @xmath340 has exactly one reeb chord , of length determined by @xmath105 .
consider the product @xmath345 of a single 1-sphere and a single 2-sphere .
since the parities of the dimensions are different , the dimension of the product is odd and the thurston - bennequin invariant will be useful .
we have @xmath346 denote the reeb chords @xmath347 and @xmath348 .
the @xmath349 factor for this pair is @xmath350 thus @xmath351 now , consider the product @xmath352 of a single 1-sphere and a single 4-sphere .
we have @xmath353 the @xmath349 factor for this pair is @xmath354 thus @xmath355 let @xmath357 be three whitney unknots . to calculate the thurston - bennequin invariant of @xmath358
apply theorem [ thrm : tb ] twice .
the torus @xmath359 has @xmath360 and @xmath361 and , after perturbing , has six reeb chords with actions @xmath362 @xmath363 @xmath364 @xmath365 and signs @xmath366 @xmath367 where @xmath368 is a unique morse critical point of index @xmath369 . since @xmath370 the only potential nonzero term in theorem [ thrm : tb ] is the @xmath349-term when computing @xmath44 for the triple product .
note that for the a chords , @xmath371 , so that @xmath372 since the signs of the morse critical points @xmath373 cancel .
similarly , the @xmath349 terms involving the b chords cancel .
now , @xmath374
@xmath375 since @xmath376 and @xmath377 .
thus , @xmath378 in other words , @xmath379 if @xmath380 satisfy the triangle inequality and @xmath360 otherwise .
the product of two legendrian knots is a torus , whose euler characteristic and thurston - bennequin invariant are 0 .
however , there are interesting @xmath44 calculations for products of three legendrian knots .
let @xmath381 be a collection of three legendrian knots with reeb chords @xmath382 .
define @xmath383 then the arguments in example 5.2 can be repeated for each triple of reeb chords @xmath380 and the thurston - bennequin invariant is @xmath384 take @xmath385 to be a once - stabilized unknot , @xmath386 a standard unknot after applying a reidemeister-1 move , and @xmath387 a right - handed trefoil .
these can be chosen to have the front projections and lagrangian projections depicted in figure 1 .
@xmath385 has two reeb chords @xmath398 ; @xmath386 has three reeb chords @xmath399 ; and @xmath387 has five reeb chords @xmath400 . by abuse of notation , @xmath401 will refer to both the chord and its action .
these chords have signs @xmath402 so @xmath403 , @xmath404 and @xmath405 .
the reeb chord actions can not be chosen completely arbitrarily as each face of the knot diagram in the lagrangian projection must satisfy the area identity determined by stokes s theorem . for a face @xmath406 of @xmath27 ,
the boundary @xmath407 lies in @xmath408 and the double points of the knot projection split the boundary into segments @xmath409 , indexed counter - clockwise .
a corner of the face is _ positive _ if near the reeb chord , the @xmath89-coordinate of @xmath410 is greater than the @xmath89-coordinate of @xmath411 and _ negative _ otherwise .
( area identity ) let @xmath412 be a face of the knot diagram in the lagrangian projection and @xmath413 be its boundary . then @xmath414 where @xmath415 are the reeb chords corresponding to the positive and negative corners of @xmath412 since each face of a knot diagram must have positive area , it follows that @xmath416 and that this is the only restriction on the actions . with this in mind and without loss of generality ,
let @xmath417 refer to some chord in the sets @xmath418 , respectively .
the minimum @xmath44 is achieved if all triples @xmath419 satisfy the triangle inequality and none of the triples @xmath420 satisfy it , since @xmath421 and @xmath422 .
then @xmath423 for instance , set @xmath424 , @xmath425 and @xmath426 .
now , for a 5-tuple @xmath427 , if both triples @xmath420 satisfy the triangle inequality , then @xmath428 must as well since @xmath429 and @xmath430 .
however , this is not true if only one such triple does .
thus , since there are more @xmath431 chords than @xmath432 chords , the maximum @xmath44 is achieved if all possible triples @xmath433 satifies the triangle inequality but no other triple does . for instance , set @xmath424 , @xmath434 , @xmath435 and @xmath436
. then @xmath437 by switching the values assigned to the @xmath348 s and @xmath105 s , it s possible to achieve many @xmath44 values between @xmath438 and @xmath439 .
s. tabachnikov .
an invariant of a submanifold that is transversal to a distribution , ( russian ) .
_ uspekhi mat .
volume 43 , number 3(261 ) , ( 1988 ) , 193194 .
translation in russian math .
surveys 43 ( 1988 ) , no .
3 , 225226 . | this paper introduces two constructions of legendrian submanifolds of @xmath0 , called legendrian products and spinning , and computes their classical invariants , the thurston - bennequin invariant and the maslov class .
these constructions take two legendrians @xmath1 and returns a product @xmath2 , they generalize other previous constructions in contact topology , such as frontspinning and hypercube tori , and are equivalent in @xmath3 .
interestingly , this construction relies upon the explicit embeddings of @xmath1 and not their legendrian isotopy class . |
[ [ section ] ] rational cherednik algebras were introduced by etingof and ginzburg in @xmath3 . since their introduction
they have been extensively studied and have been shown to be related to many other branches of mathematics such as integrable systems , symplectic algebraic geometry and algebraic combinatorics . in this article
we continue the study , initiated in @xcite , @xcite , @xcite and @xcite , of these algebras at @xmath0 and over a field of positive characteristic .
we focus on the representation theoretic aspects of the story .
in particular , we examine the block structure of certain finite - dimensional quotient algebras called restricted rational cherednik algebras .
we also look at the question of when the centre of the rational cherednik algebra is smooth .
analogous problems have already been solved for rational cherednik algebras at @xmath4 in characteristic zero , see @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , and our results are very similar in nature .
the methods we develop , however , are new , and in fact can be used to reprove many of the characteristic zero results . [ [ ss : maintheorem1 ] ] let us review our results .
further details can be found in the main body of the paper .
let @xmath5 be a pseudo - reflection group .
let @xmath6 be an algebraically closed field of characteristic @xmath7 with @xmath8 .
let @xmath9 denote the reflection representation of @xmath5 over @xmath6 .
let @xmath10 denote the set of reflections in @xmath5 and let @xmath11 be a @xmath5-invariant function . to this data
we can attach a @xmath6-algebra @xmath12 called the _ rational cherednik algebra_. let @xmath13 denote the frobenius twist of @xmath9 .
let @xmath14 denote the centre of @xmath12 .
there is an injective algebra homomorphism @xmath15^w \otimes k[(v^*)^{(1)}]^w \hookrightarrow z_{\bc}(w).\ ] ] factoring out by the unique graded , maximal ideal in this central subalgebra , one gets a finite dimensional , graded quotient of the rational cherednik algebra .
this factor algebra is called the restricted rational cherednik algebra and is denoted @xmath16 .
simple modules for this finite dimensional algebra are in natural bijection with the simple modules of the group @xmath5 .
thus , the blocks of @xmath16 give us a partition of the set @xmath17 .
our first main result is an explicit combinatorial description of this block partition when @xmath5 belongs to the infinite series @xmath18 . in this case , the irreducible representations of @xmath5 are naturally labeled by @xmath19 , the set of @xmath20-multipartitions of @xmath21 . for each @xmath22 , let @xmath23 denote the corresponding simple @xmath16-module
. for the definitions in the following statement and the proof of the following theorem , see [ blocks ] .
let @xmath24 .
let @xmath25 .
then @xmath23 and @xmath26 belong to the same block of @xmath16 if and only if @xmath27 using this description of the blocks of the restricted rational cherednik algebra @xmath28 , some clifford theory also allows one to describe the blocks of the restricted rational cherednik algebra @xmath29 , see [ ss : g(m , d , n)blocks ] . here
@xmath1 denotes ( in the shepard - todd classification ) the normal subgroup of @xmath18 where we impose the restriction that @xmath30 and either @xmath31 or @xmath32 and @xmath33 is odd .
[ [ smooth+blocks ] ] our remaining results concern the smoothness of @xmath14 . in section
[ sec : smoothness ] we relate the smoothness of @xmath14 to the representation theory of @xmath12 .
taking as our starting point the fact that the smooth and azumaya loci of @xmath14 are equal , we use the restriction functors of @xcite to establish the following .
the following are equivalent : 1 .
@xmath14 is smooth ; 2 .
the blocks of @xmath34 are singletons for all parabolic subgroups @xmath35 . here
, @xmath36 denotes the restriction of @xmath37 to @xmath38 . in the case
that @xmath39 , we can apply the theorem together with the description of blocks from theorem [ ss : maintheorem1 ] to determine for which parameters @xmath37 the centre @xmath14 is smooth .
[ cor : gmonensmooth ] the centre of @xmath40 is smooth if and only if @xmath37 does not lie on the finitely many hyperplanes in @xmath41 defined by @xmath42 we should note that an analogous version of the theorem and its corollary is true for rational cherednik algebras at @xmath43 over the complex numbers . using ( * ? ? ?
* theorem 5.5 ) , this gives an alternative way to describe the parameter values where the centre is smooth , cf .
* lemma 4.3 ) .
the theorem also clarifies the relationship between restricted cherednik algebras and the smoothness of the centres @xmath14 . [
[ genericsmooth ] ] the corollary shows that the centre of @xmath40 is a smooth algebra for generic values of the parameter @xmath37 .
one can ask more generally : for which pseudo - reflection groups @xmath5 is @xmath14 smooth for generic values of the parameter @xmath37 ?
our final result answers this question .
[ thm : classify ] the centre of the rational cherednik algebra @xmath12 is never smooth if @xmath5 is not isomorphic to @xmath18 or @xmath44 . if @xmath5 is isomorphic to @xmath18 for some @xmath20 and @xmath21 or to the exceptional group @xmath44 , then the centre of @xmath12 is smooth for generic values of @xmath37 . our approach to proving theorem [ thm : classify ] follows the same path as in @xcite - we calculate the poincar polynomial of the graded @xmath16-modules @xmath23 under the assumption that the dimension of these modules is maximal , namely @xmath45 .
this leads to a contradiction for many choices of @xmath46 .
however , the naive argument of is not sufficient in our case to calculate this poincar polynomial .
instead , we show that @xmath23 can be deformed ( flatly ) to a graded @xmath47-module . using a result of cartier s on @xmath48-modules with zero @xmath7-curvature
, we study the graded @xmath5-character of this module at @xmath49 . theorem [ thm : classify ] and corollary [ ss : maintheorem1 ] provide a complete answer to ( * ? ? ?
* question d ) for rational cherednik algebras .
[ [ section-1 ] ] the paper is structured as follows . in section 2
we introduce notation and recall some facts about pseudo - reflection groups . in section 3
we define rational cherednik algebras and state their main properties .
the dunkl - opdam operators are introduced in section 4 and are used to prove theorem [ ss : maintheorem1 ] .
section 5 is devoted to parabolic restriction and induction and their application in the proof of theorem [ smooth+blocks ] .
finally , in section 6 we establish certain properties of @xmath48-modules in characteristic @xmath7 and use these to prove theorem [ genericsmooth ] .
the first author is supported by the epsrc grant ep - h028153 .
the second author was supported by the sfb / tr 45 periods , moduli spaces and arithmetic of algebraic varieties " of the dfg ( german research foundation ) .
the authors would like to thank ulrich thiel for suggesting that the euler element should distinguish the blocks for @xmath44 at generic parameters , and for showing us an early version of @xcite .
let @xmath6 be an algebraically closed field of characteristic @xmath50 .
let @xmath9 be a @xmath6-vector space of finite dimension and @xmath5 a finite group acting linearly on @xmath9 .
an element @xmath51 is called a pseudo - reflection if the fixed space of @xmath52 has co - dimension one .
let @xmath53 denote the set of all pseudo - reflections in @xmath5 .
then @xmath5 is said to be a if @xmath54 , a good reference on the theory of pseudo - reflection groups is @xcite .
we assume throughout that @xmath55 does * not * divide @xmath56 .
this assumption on the characteristic of @xmath6 implies that there are no transvections in @xmath53 .
one can also check from the classification of pseudo - reflection groups , as recalled in @xcite , that this assumption implies that @xmath5 is the reduction mod @xmath7 of a complex reflection group .
let @xmath9 be a finite dimensional vector space over @xmath6 .
let @xmath5 be a finite group acting linearly on @xmath9 and assume that @xmath7 does not divide @xmath56 .
the frobenius morphism @xmath57 \ra k[v]$ ] is the ring homomorphism @xmath58 .
denote by @xmath59 $ ] the image of @xmath60 but with twisted linear structure @xmath61 for all @xmath62 $ ] and @xmath63 .
then @xmath64 \ra k[v^{(1)}]$ ] is a @xmath6-linear isomorphism .
note that @xmath65 $ ] is a finite free @xmath59$]-module of rank @xmath66 .
it is easy to check that @xmath60 is a @xmath5-equivariant ring homomorphism , so we have an isomorphism @xmath67^w \ra k[v^{(1)}]^w$ ] . in particular , the @xmath7-th powers of a generating set for @xmath65^w$ ]
form a generating set for @xmath59^w$ ] .
let us retain the notation from above .
let @xmath68 be a finite field extension of @xmath69 containing all the @xmath56th roots of unity , let @xmath70 be the localisation of the ring of integers of @xmath68 at the prime ideal generated by @xmath71 and let @xmath72 be the residue field of @xmath70 .
note that @xmath72 is a finite field of order a power of @xmath7 . with this setup we can define a decomposition map on characters of irreducible representations
as follows , see ( * ? ? ?
* @xmath73 7 ) .
let @xmath74 be an irreducible @xmath75-module .
we can choose an @xmath70-lattice @xmath76 in @xmath74 so that the action of @xmath5 on @xmath76 has structure constants in @xmath70 .
let @xmath77 denote the reduction of @xmath76 to @xmath72 , which is naturally a @xmath78-module .
the decomposition map is the assignment @xmath79 , where @xmath80 denote the characters of @xmath74 and @xmath77 , respectively . by (
* corollary 17.2 ) and tits deformation theorem , ( * ? ? ?
* theorem 7.4.6 ) , we have the following . both @xmath75 and @xmath78
are split algebras and the decomposition map defines a bijection between the irreducible characters of @xmath75 and @xmath78 .
in particular , it follows that the irreducible characters of @xmath81 are given by reducing the irreducible characters of @xmath82 to @xmath6 .
let @xmath5 be a pseudo - reflection group and let @xmath83 denote the grothendieck group of finite dimensional @xmath5-modules .
let @xmath17 be a complete set of isomorphism classes of simple @xmath5-modules .
we denote by @xmath84 a complete set of isomorphism classes of graded , simple @xmath5-modules . for a graded @xmath5-module @xmath74
, we write @xmath85 $ ] for its graded character .
the shift @xmath86 $ ] of @xmath74 is the graded @xmath5-module such that @xmath86_j = m_{j - i}$ ] .
we endow the algebra @xmath65 $ ] with its usual @xmath87-grading .
let @xmath88 be the degrees of a set of fundamental homogeneous algebraically independent generators of @xmath65^w$ ] .
the coinvariant ring of @xmath5 is @xmath65^{\mathrm{co } w}$ ] , and for each @xmath89 , we denote by @xmath90 the corresponding fake polynomial , defined by @xmath91^{\mathrm{co } w } ) = \sum_{\lambda \in \irr w } f_{\lambda}(t ) \cdot [ \lambda].\ ] ] let @xmath59^w_+$ ] denote the invariant polynomials with zero constant term .
the ring is defined to be the finite dimensional graded algebra @xmath92 / \
< k[v^{(1)}]^w_+ \>$ ] .
[ lem : pcochar ] keep notation as above .
then there is an isomorphism of graded @xmath5-modules @xmath93 / \ <
k[v]^w_+ \ > ) \o ( k[v]^w / \
< k[v^{(1)}]^w_+ \ > ) \\ & \simeq ( k[v ] / \
< k[v^{(1)}]_+ \ > ) \o ( k[v^{(1 ) } ] / \
< k[v^{(1)}]^w_+ \>).\end{aligned}\ ] ] consider the inclusions of algebras @xmath94^w \ar[dr ] & \\ k[v^{(1)}]^w \ar[ur ] \ar[dr ] & & k[v ] \\ & k[v^{(1 ) } ] \ar[ur ] & } \ ] ] choose a basis @xmath95 of @xmath96 , homogeneous generators @xmath97 of @xmath65^w$ ] and a free homogeneous basis @xmath98 of @xmath65 $ ] as a @xmath65^w$]-module
. then there are free bases for @xmath65 $ ] as a @xmath59^w$]-module given by : @xmath99 and @xmath100 ) implies that @xmath101t^i = f_\lambda(t ) \cdot \prod_{i = 1}^n \frac{1 - t^{p d_i}}{1 - t^{d_i}}\ ] ] for all @xmath89 .
[ [ sec : defncherednik ] ] for @xmath102 , fix @xmath103 to be a basis of the one dimensional space @xmath104 and @xmath105 a basis of the one dimensional space @xmath106 , normalised so that @xmath107 .
let @xmath41 denote the space of @xmath5-equivariant functions @xmath108 and choose @xmath109 and @xmath110 .
the _ rational cherednik algebra _ , @xmath111 , as introduced by etingof and ginzburg @xcite , is the quotient of the skew group algebra of the tensor algebra , @xmath112 , by the ideal generated by the relations @xmath113 = 0 , \qquad [ y , y ' ] = 0 , \qquad [ y , x ] = t x(y ) - \sum_{s \in \mathcal{s } } \bc(s ) \alpha_s(y ) x(\alpha_s^\vee ) s,\ ] ] for all @xmath114 and @xmath115 .
we define a filtration @xmath116 on @xmath117 via @xmath118 , @xmath119 and @xmath120 for @xmath121 . by (
* theorem 1.3 ) , there is an isomorphism of algebras @xmath122 as a consequence , there is a vector space isomorphism @xmath123 \otimes k w \otimes k [ v^*].\ ] ] there is also a @xmath124-grading on @xmath12 given by setting @xmath125 , @xmath126 and @xmath127 . throughout this article
we assume that @xmath128 .
therefore , without loss of generality @xmath129 and we write @xmath130 for @xmath131 .
let @xmath132 be a basis of @xmath96 and @xmath133 the dual basis .
define the euler element in @xmath130 to be @xmath134 one can easily check that @xmath135 = x$ ] , @xmath136 = -y$ ] and @xmath137 = 0 $ ] for all @xmath138 , @xmath139 and @xmath140
. therefore the element @xmath141 belongs to the centre of @xmath12 .
[ [ cherednik_props ] ] below we summarize fundamental the properties of @xmath12
. proofs of all these statements can be found in @xcite .
let @xmath142 be a rational cherednik algebra associated to @xmath143 . 1 .
the p.i . degree of @xmath144 equals @xmath145 .
the centre @xmath14 of @xmath144 is an affine domain and the algebra @xmath144 is a finite module over its centre .
3 . the smooth locus of @xmath14 equals the azumaya locus of @xmath144 .
the commutative subalgebras @xmath59^w$ ] and @xmath146^w$ ] of @xmath144 are central .
let @xmath72 be a simple , graded @xmath12-module .
the centre of @xmath12 acts as a scalar on @xmath72 and the grading on @xmath72 implies that @xmath147^w \o k[(v^*)^{(1)}]^w)_+$ ] annihilates @xmath72 . therefore to study these simple
, graded modules it suffices to consider a certain graded , finite dimensional quotient of @xmath12 .
the @xmath148 is the finite dimensional quotient of @xmath12 by the central ideal generated by @xmath147^w \o k[(v^*)^{(1)}]^w)_+$ ] .
the algebra @xmath149 is @xmath124-graded and has dimension @xmath150 . the pbw property ( [ eq : pbw ] ) implies that @xmath151^{p \mathrm{co } w } \otimes k w \otimes k [ v^*]^{p \mathrm{co } w}\ ] ] as vector spaces . since @xmath152 is a complete intersection , ( * ? ? ?
* corollary 21.19 ) implies that it is gorenstein and thus equipped with a non - degenerate bilinear form .
then the proof of ( * ? ? ?
* theorem 3.6 ) shows that the algebra @xmath148 is symmetric .
let @xmath153 .
the @xmath154 associated to @xmath155 is the induced module @xmath156 where @xmath157^{\mathrm{pco } w } \rtimes w$ ] and the natural action of @xmath5 on @xmath155 extends to @xmath70 by making @xmath9 act by zero . as was done in @xcite
, we can apply the theory developed in @xcite to the category @xmath158 of finite - dimensional , graded , left @xmath148-modules .
the forgetful functor @xmath159 is denoted @xmath160 .
[ prop : listofproperties ] let @xmath161 . 1 .
the baby verma module @xmath154 has a simple head @xmath162 .
@xmath154 is isomorphic to @xmath163 if and only if @xmath164 in @xmath84 .
the set @xmath165 is a complete set of isomorphism classes of simple modules in @xmath158 .
4 . the set @xmath166 is a complete set of isomorphism classes of simple @xmath148-modules .
proposition [ prop : listofproperties ] shows that there is a natural bijection , @xmath167 , between @xmath17 and @xmath168 .
therefore the blocks of @xmath148 define a partition , which we call the _ block partition _
, of the set @xmath17 .
[ [ wreathcherednik ] ] let @xmath169 be integers . let @xmath170 be the cyclic group of order @xmath20 .
we fix a generator @xmath171 and let @xmath172 denote the transposition which swaps @xmath173 and @xmath174 .
we denote by @xmath175 the simple transposition swapping @xmath174 and @xmath176 . the group @xmath39 is the semidirect product @xmath177 .
we write @xmath178 for the element @xmath179 with @xmath180 in the @xmath173th place .
let @xmath181 be the reflection representation of @xmath18 . recall that we assume that @xmath8 , so in particular @xmath182 .
let @xmath183 be a primitive @xmath20th root of unity .
we fix a basis @xmath184 of @xmath9 so that @xmath185 for all @xmath186 and all @xmath187 .
let @xmath188 be the dual basis .
the conjugacy classes of reflections in @xmath5 are given by @xmath189 and , for each @xmath190 , @xmath191 the parameter @xmath192 is represented by @xmath193 , where @xmath194 and @xmath195 .
using this notation , the definition of the rational cherednik algebra becomes the following .
let @xmath39 .
then @xmath12 is the quotient of @xmath112 by the relations : @xmath196 = 0,\ \ [ y_i , y_j]=0 , \\ & [ y_i , x_i ] = 1 - { \kappa } \sum_{l=0}^{m-1 } \sum_{j \neq i } s_{ij } g_i^{-l } g_j^l - \sum_{l=1}^{m-1 } { c}_l(1-\eta^{-l } ) g_i^l,\\ & [ y_i , x_j ] = { \kappa}\sum_{l=0}^{m-1 } \eta^{-l } s_{ij } g_i^{-l } g_j^{l}.\end{aligned}\ ] ] for all @xmath197 , we define elements in @xmath12 : @xmath198 by ( * ? ? ?
* lemma 3.2 ) , @xmath199=0 $ ] for all @xmath200 .
let @xmath201 $ ] denote the algebra generated by the @xmath202 . by ( [ pbw ] ) ,
this is a polynomial algebra .
we denote by @xmath203 and @xmath204 the @xmath205th elementary symmetric polynomial and @xmath205th power sum in the @xmath206 , respectively . by convention ,
we will use the following result from @xcite . keep notation as above . then @xmath208 = \sum_{1 < j_2 < \dots < j_r \leq n } x_1
z_{j_2 } \dots z_{j_r}.\ ] ] [ [ power_sums ] ] let @xmath209 and @xmath210
. in this notation theorem [ doops ] reads @xmath211 = x_1 e_r'.\end{aligned}\ ] ] recall newton s formula : @xmath212 applying @xmath213 to ( [ newton ] ) we get @xmath214 a straightforward induction argument using the fact that @xmath215 shows that @xmath216 we will need a simple preparatory lemma .
let @xmath217 .
we have @xmath218 the proof is by induction .
the case @xmath219 is clear . for @xmath220 , we have @xmath221 and the induction step follows by a simple calculation which we leave to the reader .
for @xmath222 , @xmath223= x_1 q_r.\ ] ] the proof goes by induction on @xmath205 .
for @xmath224 we have @xmath225 and @xmath226 . by theorem [ doops ] ,
@xmath227=x_1 $ ] .
. then @xmath229 & = [ re_r + \sum_{i=1}^{r-1 } ( -1)^{i}p_ie_{r - i},x_1 ] \notag \\ \label{eqn1 } & = r x_1 e_r ' + \sum_{i=1}^{r-1 } ( -1)^{i } ( p_ix_1e_{r - i } ' + x_1 q_i e_{r - i}).\end{aligned}\ ] ] the first line follows from rewriting @xmath204 using ( [ newton ] ) , and the second line follows from the induction hypothesis and ( [ comm_symm ] ) . now ( [ eqn1 ] ) equals @xmath230\\ \notag = & x_1 \left [ \sum_{i=1}^r ( -1)^{i-1}p_i ' e_{r - i } + \sum_{i=1}^{r-1 } ( -1)^{i } ( q_i(e_{r - i } ' + e_{r - i } ) \right]\\ \label{eqn2 } = & x_1 \left [ \sum_{i=1}^r ( -1)^{i-1}iz_1^{i-1 } e_{r - i } + \sum_{i=1}^{r-1 } ( -1)^{i } ( q_i(\sum_{k=0}^{r - i-1 } ( -1)^{r - i - k-1 } z_1^{r - i - k-1}e_{k } ) + q_ie_{r - i } ) \right].\end{aligned}\ ] ] here the first line follows from the induction hypothesis , the second from ( [ derive_e ] ) and the fourth from ( [ derive_e2 ] ) . for a fixed @xmath231 ,
the coefficient of @xmath232 in ( [ eqn2 ] ) is equal to @xmath233 which equals zero by the lemma above . by the lemma above
, the coefficient of @xmath234 is @xmath235 therefore , @xmath223 = x_1 q_r.\ ] ] [ [ power_sums_central ] ] we can now provide some central elements in @xmath130 . for all @xmath236 , @xmath237 thus , @xmath238^{s_n } \subset z_{\bc}(w).$ ]
we continue to write @xmath239 for the power sum in the variables @xmath240 .
we have @xmath241 therefore by proposition [ power_sums ] , @xmath242 & = x_1 \sum_{j=0}^r { r\choose j } ( -1)^{r - j } q_{pj+r - j}\\ & = x_1 \sum_{j=0}^r { r\choose j } ( -1)^{r - j } ( ( z_1 + 1)^{pj+r - j } - z_1^{pj+r - j})\\ & = x_1 \sum_{j=0}^r { r\choose j } ( -1)^{r - j } ( ( z_1^p+1)^j ( z_1 + 1)^{r - j } - ( z_1^p)^jz_1^{r - j})\\ & = x_1\left [ ( z_1^p+1 ) - ( z_1 + 1 ) \right]^r - x_1 ( z_1^p - z_1)^r = 0 . \end{aligned}\ ] ] the theorem now follows as in ( * ? ? ?
* theorem 3.4 ) .
the element @xmath243 is symmetric in the @xmath202 so it commutes with any @xmath187 , see ( * ? ? ?
* lemma 5.1 ) .
therefore this power sum commutes with @xmath244 for all @xmath245 , and in particular with each @xmath246 . to see that the power sum also commutes with the @xmath247 , we use the isomorphism @xmath248 from @xcite , where @xmath249 is also defined . by definition @xmath250 and @xmath251 , so that @xmath252 for all @xmath173 . applying @xmath253 to @xmath254 $ ] shows that @xmath255=0 $ ] for all @xmath173 .
we use an identical argument to the proof of ( * ? ? ?
* theorem 5.5 ) , to determine the blocks of @xmath256 .
we use freely the notation from ( * ? ? ?
* @xmath735 ) .
let us first change our parameters .
we define @xmath257 via @xmath258 we denote by @xmath19 the set of @xmath20-multipartitions of @xmath21 , @xmath259 .
the simple representations of @xmath81 are labeled by the set @xmath19 ; over the complex numbers this is standard and the same holds for @xmath81 by reduction , cf .
[ repsreduction ] .
in fact the construction from @xcite is also valid over @xmath6 , since @xmath7 does not divide @xmath56 , so that we can find bases of irreducible @xmath81-modules given by eigenvectors of jucys - murphy elements . given a partition @xmath155 , the residue @xmath260 is an element in @xmath261 $ ] , the group ring of the additive group @xmath262 . for @xmath263 and @xmath264 ,
let @xmath265 .
let @xmath24 .
let @xmath266 .
then @xmath23 and @xmath26 belong to the same block of @xmath16 if and only if @xmath267 since this follows the proof of ( * ? ? ?
* theorem 5.5 ) closely , we shall sketch the argument , pointing out the necessary changes to using theorem [ power_sums_central ] together with weyl s theorem for invariants of symmetric groups ( which is valid whenever @xmath8 , see ( * ? ? ?
* theorem 3.3.1 ) ) , it is enough to calculate the characters of @xmath243 on each @xmath23 . the characters are determined by choosing a simultaneous eigenvector @xmath268 for @xmath240 .
the eigenvalues for @xmath269 are evaluated as in @xcite , and produce the desired combinatorial description .
the blocks of @xmath256 for @xmath270 , where @xmath30 and either @xmath31 or @xmath32 and @xmath33 is odd , can be calculated from theorem [ blocks ] by using clifford theory as in @xcite .
the resulting description of the blocks is then analogous to that given in characteristic zero , see @xcite .
let @xmath271 . let @xmath272 denote the set @xmath273 .
let @xmath274 be the algebra of crystalline differential operators on @xmath272 .
for @xmath275 , let @xmath276 denote the dunkl operator @xmath277 .
by @xcite there is an injective algebra morphism @xmath278 for all @xmath140 , @xmath279 and @xmath275 .
this embedding becomes an isomorphism after localizing @xmath12 at the ore set @xmath280 .
let us recall the setup of @xcite .
let @xmath281 be the stabilizer of a point @xmath282 and let @xmath283 .
for any point @xmath282 we write @xmath284_b$ ] for the completion of @xmath65 $ ] at @xmath285 , and we write @xmath286}_b$ ] for the completion of @xmath65 $ ] at the @xmath5-orbit of @xmath285 in @xmath9 .
note that we have @xmath284_0=\widehat{k[v]}_0 $ ] .
for any finitely generated @xmath65$]-module @xmath74 , let @xmath287}_b\otimes_{k[v ] } m.\ ] ] the completion @xmath288 is the algebra generated by @xmath289}_b$ ] , the dunkl operators @xmath276 for @xmath275 , and the group @xmath5 .
let @xmath249 denote the restriction of @xmath37 to @xmath290 .
the algebra @xmath291 is then defined similarly .
let @xmath292 be the set of @xmath293-equivariant maps from @xmath5 to @xmath294 .
let @xmath295 be the ring of endomorphisms of the right @xmath294-module @xmath296 .
the following proposition is proved over @xmath297 in ( * ? ? ?
* theorem 3.2 ) , and has an identical proof over @xmath6 .
[ beiso ] there is an isomorphism of algebras @xmath298 defined as follows : for @xmath299 , @xmath300 , @xmath301 , @xmath302 , @xmath303 where @xmath304 , @xmath305 , @xmath306 and @xmath307 .
let @xmath308 be defined by @xmath309 if @xmath310 and @xmath311 otherwise .
then @xmath312 is a primitive idempotent and @xmath313 .
the bimodule @xmath314 yields a morita equivalence between @xmath291 and @xmath315 .
[ [ ss : small_modules ] ] suppose that @xmath74 is a finite dimensional @xmath316-module and let @xmath317 denote the support of @xmath74 as a @xmath65^w$]-module .
if @xmath317 is the @xmath5-orbit of some point @xmath318 , then @xmath74 is naturally a @xmath319-module . in particular
, the support of any simple @xmath316-module is a single @xmath5-orbit and so is a @xmath288-module . let @xmath74 be a @xmath288-module , then denote by @xmath320 the @xmath315-module given by transporting the @xmath288-action on @xmath74 via @xmath321 .
an @xmath322-module @xmath74 is called _ small _ if @xmath323 .
[ prop : completesmall ] a @xmath316-module @xmath74 is small if and only if @xmath324 is a small @xmath325-module . note first that @xmath324 is a @xmath325-module by restricting the @xmath326-action .
let @xmath327 be right coset representatives of @xmath293 in @xmath5 .
then @xmath328 , with @xmath329 for all @xmath200 .
in particular , @xmath323 if and only if @xmath330 .
let @xmath331 denote the centre of @xmath316 .
keep notation as above .
then @xmath331 is smooth if and only if @xmath325 has no small modules @xmath74 with @xmath332 for all parabolic subgroups @xmath35 . by proposition [ cherednik_props ]
( 3 ) , @xmath331 is smooth if and only if @xmath316 has no small modules .
let @xmath74 be a small , simple @xmath316-module and choose @xmath318 such that the support of @xmath74 equals the @xmath5-orbit of @xmath285 .
then @xmath74 extends to a @xmath288-module and proposition [ prop : completesmall ] says that @xmath324 is a small @xmath325-module , where @xmath293 is the stabilizer of @xmath285 in @xmath5 .
moreover , the support of @xmath324 is @xmath333 .
conversely , if there exists a small @xmath325-module @xmath334 supported on @xmath333 then this extends to a @xmath291-module .
since @xmath291 is morita equivalent to @xmath335 , there exists some @xmath336-module @xmath74 such that @xmath337 . proposition [ prop : completesmall ] says that @xmath74 is actually a small @xmath316-module .
[ [ section-2 ] ] let @xmath74 be a simple @xmath316-module with @xmath338 .
we can define an induced module ( cf .
[ rrca ] ) as follows .
let @xmath339 \rtimes w$ ] and let @xmath340 .
extend the @xmath5-action on @xmath341 to a @xmath342-action by letting @xmath343 $ ] act by @xmath344 .
define @xmath345 by the support condition on @xmath74 , the subspace @xmath346 of @xmath74 is non - zero . it is a @xmath5-submodule of @xmath74 and we may assume without loss of generality that @xmath347 . therefore there is a surjective homomorphism @xmath348 , which maps @xmath349 to @xmath347 in the obvious way . recall that @xmath316 has a @xmath124-grading .
we make @xmath341 into a graded @xmath342-module by setting @xmath350 and @xmath351 for all @xmath352 .
we give @xmath353 a @xmath124-graded module structure by inducing the graded structure on @xmath341 .
since @xmath74 is simple , there is some @xmath354 such that @xmath355 , where @xmath356 denotes the maximal ideal corresponding to @xmath357 .
define @xmath358 by our choice of @xmath357 , there is also a surjective homomorphism @xmath359 .
the grading on @xmath353 induces filtrations on @xmath360 and @xmath74 .
we use the notation @xmath361 to denote the associated graded objects with respect to these filtrations .
recall that a morphism @xmath362 between filtered @xmath316-modules is called a morphism if @xmath363 for all @xmath364 .
the functor @xmath361 is exact on short exact sequences of strictly filtered morphisms .
note also that the surjection @xmath365 is strictly filtered by definition .
the filtration on @xmath353 is both exhaustive and separating , therefore the same is true of @xmath360 .
however , this module is finite dimensional therefore we have @xmath366 and @xmath367 for @xmath368 . [
prop : assocgraded ] the @xmath316-module @xmath369 has dimension @xmath370 and is annihilated by @xmath147^w \o k[(v^*)^{(1)}]^w)_+$ ] .
denote by @xmath371 the @xmath173th piece of the filtration on @xmath74 . by construction , @xmath372 is a decreasing filtration with @xmath373 and @xmath374 for all @xmath368 .
thus @xmath375 .
for the second statement , we show that @xmath376 .
the claim then follows since there is a surjective map @xmath377 .
consider the short exact sequence : @xmath378 where the left - hand term is given the induced filtration . by the pbw theorem and the nullstellensatz , @xmath379^w_+ \cdot ( \h_{\bc}(w , v ) \o_{b } e)$ ] . since @xmath353 is graded , @xmath380 .
thus @xmath376 .
[ cor : equivsmooth ] the following are equivalent : 1 .
@xmath331 is smooth ; 2 .
@xmath381 has no small modules @xmath74 with @xmath332 for all parabolic subgroups @xmath35 ; 3 .
@xmath382 has no small modules @xmath74 for all parabolic subgroups @xmath35 ; 4 .
the blocks of @xmath382 are singletons for all parabolic subgroups @xmath35 .
there are isomorphisms of algebras @xmath383 and @xmath384 , where @xmath385 \o k[((v^{w'})^*)^{(1)}])_+ \rangle$ ] .
in particular , ( 2 ) and ( 3 ) are equivalent to * @xmath386 has no small modules @xmath74 with @xmath332 for all parabolic subgroups @xmath35 ; * @xmath387 has no small modules @xmath74 for all parabolic subgroups @xmath35 , respectively , where we have used the fact that @xmath388 is an azumaya algebra .
the equivalence of ( 1 ) and ( 2 ) is corollary [ ss : small_modules ] .
clearly ( 2 ) implies ( 3 ) . for the converse , suppose that @xmath325 has a small module @xmath74 with @xmath332 for some parabolic subgroup @xmath35 . then by the proposition
, @xmath369 is a small module for @xmath389 .
the equivalence of ( 3 ) and ( 4 ) follows from an identical argument to @xcite .
[ rem : remark1 ] the proofs above are valid , mutando mutandae , for @xmath43 and char @xmath390 .
let @xmath391 be positive integers and assume that @xmath392 .
let @xmath393 . in this section
we give a proof of corollary [ cor : gmonensmooth ] .
the idea is to use corollary [ cor : equivsmooth ] ( 4 ) together with the results of section [ sec : blocksg(m,1,n ) ] to determine for which @xmath37 the centre @xmath14 is smooth .
recall that the parabolic subgroups of @xmath5 are of the form @xmath394 here @xmath395 denotes the symmetric group on @xmath396 letters , and @xmath397 by definition .
for such a parabolic subgroup , the representation @xmath398 is the reflection representation .
recall that @xmath399 .
recall the parameters @xmath400 from theorem [ blocks ] .
let @xmath401 denote the set of all @xmath37 such that either @xmath402 for some @xmath403 and integer @xmath404 such that @xmath405 .
thus @xmath401 is a finite union of hyperplanes in @xmath406 .
the algebra @xmath14 is smooth if and only if @xmath407 .
let us first suppose that @xmath408 . recall that , for @xmath22 , @xmath265 .
suppose that @xmath409 .
then @xmath410 and so @xmath411 .
thus by theorem [ blocks ] , @xmath412 has a non - singleton block , and so @xmath14 is singular , corollary [ cor : equivsmooth ] . if @xmath413 and @xmath414 for some @xmath405 , then @xmath415 without loss of generality @xmath416 and @xmath417 , and then @xmath418 thus @xmath412 has a non - singleton block and @xmath14 is singular .
a similar argument applies in the case @xmath419 .
suppose now that @xmath407 .
we first show that @xmath412 has only singleton blocks . for
a contradiction , suppose that there exist distinct @xmath420 such that @xmath421 . since @xmath422 , @xmath423 .
each box @xmath285 in the young diagram of @xmath424 contributes @xmath425 to @xmath426 . since @xmath427
, there exist @xmath428 and boxes @xmath429 , @xmath430 such that @xmath431 .
now @xmath432 and @xmath433 are integers such that @xmath434 and @xmath435 .
thus @xmath436 for some @xmath437 .
this means that @xmath438 , and so @xmath408 , a contradiction .
we now want to prove the stronger statement that @xmath407 implies that @xmath14 is smooth .
note that @xmath439 for all @xmath440 or @xmath20 , @xmath441 .
therefore @xmath407 implies that @xmath442 for all @xmath440 or @xmath20 , @xmath441 . by the description of parabolic subgroups
@xmath443 given above , the previous paragraph implies @xmath444 has only singleton blocks for all @xmath293 . by corollary [ cor : equivsmooth ]
this implies that @xmath14 is smooth .
[ rem : symsing ] 1 . in the case
@xmath445 , @xmath5 is just the symmetric group @xmath446 .
although @xmath447 is not the reflection representation , we have @xmath448 , where @xmath449 now denotes the reflection representation of @xmath446 .
the set @xmath450 is then identified with @xmath451 .
note that the proof of the above theorem shows that for @xmath39 , the centre @xmath14 is smooth if and only if @xmath412 has only singleton blocks .
in this section we describe , based on cartier s theorem , the category of @xmath452-module with @xmath7-curvature zero .
this will allow us to prove theorem [ thm : classify ] .
fix a basis @xmath132 of @xmath96 and @xmath453 of @xmath9 such that @xmath454 .
let @xmath455\quad \textrm{and } \quad \operatorname{\mathrm{spec}}a = t^{*,(1 ) } v.\ ] ] the centre of @xmath456 embeds in @xmath70 and we write @xmath457 for the corresponding finite morphism .
the group @xmath5 acts on @xmath458 and @xmath459 such that the map @xmath460 is @xmath5-equivariant and satisfies @xmath461 for fixed @xmath462 and @xmath463 , we define @xmath464 , where @xmath70 acts on @xmath155 via the character @xmath465 . 1 .
the @xmath452-module @xmath466 is simple ; 2 .
@xmath467 if and only if @xmath468 and , moreover , if @xmath469 then @xmath470 via the conjugation isomorphism @xmath471 ; 3 .
every simple @xmath452-module is isomorphic to @xmath466 for some @xmath465 and @xmath155 .
considered as an @xmath70-module , @xmath472 where @xmath473 is set - theoretically supported at @xmath474 . each @xmath475 is a @xmath476-submodule of @xmath466 and @xmath466 will be a simple @xmath452-module if and only if @xmath477 is a simple @xmath478-module . if @xmath479 with @xmath480 then we write @xmath481 with @xmath318 for the unique closed point in @xmath482 . applying the @xmath483-equivariant automorphism @xmath484 and @xmath485 to @xmath456 , we may assume without loss of generality that @xmath486 and @xmath487 .
let @xmath488 be the unique simple @xmath456-module supported at @xmath489 ; simplicity of @xmath490 follows from the fact that @xmath456 is azumaya of rank @xmath491 and the dimension of @xmath490 is @xmath492 .
then @xmath493 , with @xmath5 acting diagonally .
the module @xmath494 is simple : let @xmath495 be a basis of @xmath490 such that @xmath496 and choose any @xmath497 .
choose an @xmath173 such that @xmath498 .
since @xmath456 surjects onto @xmath499 , there is some @xmath500 such that @xmath501 for all @xmath502 and @xmath503 .
then @xmath504 and we have @xmath505 . hence @xmath506 . to show that the various @xmath507 are non - isomorphic , note that @xmath508 because the space @xmath509 is one - dimensional .
arguing geometrically as above shows the second claim of the statement .
the final statement is clear just by considering the socle , as a @xmath70-module , of an simple @xmath452-module .
now we require a special case , proposition [ prop : cartier ] , of a classical result by cartier on @xmath48-module with zero @xmath7-curvature .
we follow the presentation given in @xcite . recall that if @xmath510 is a derivation , then @xmath511 also acts as a derivation and we write @xmath512}$ ] for this derivation
so that @xmath513}$ ] acts trivially on @xmath514 $ ] . 1 .
the module @xmath74 has zero @xmath7-curvature if and only if @xmath74 is scheme - theoretically supported on @xmath13 when considered as a @xmath524-module i.e. @xmath525 for all @xmath173 .
the horizontal sections " functor @xmath526 \rtimes w\text{-}{\mathsf{mod}}}$ ] , @xmath527 is an equivalence of categories with quasi - inverse @xmath528 \rtimes w\text{-}{\mathsf{mod } } } \ra \mc{d}_0 $ ] , @xmath529 where @xmath70 acts on @xmath530 via the morphism @xmath531 $ ] , @xmath532 .
the equivalence @xmath533 restricts to an equivalence of graded categories @xmath534 \rtimes w\text{-}{\mathsf{grmod}}}$ ] . 1 .
it is shown in ( * ? ? ?
* proposition 5.2 ) that @xmath544 is @xmath7-linear i.e. @xmath545 for all @xmath546 $ ] and @xmath547 . since @xmath548 for all @xmath173 it is clearly necessary that @xmath549 , i.e. @xmath525 , in order for @xmath74 to have zero @xmath7-curvature . on the other hand ,
every element @xmath550 can be expressed as @xmath551 for some @xmath546 $ ] ; thus @xmath552 and @xmath553 if @xmath525 for all @xmath173 .
2 . as in lemma
[ lem : poperator ] , define @xmath554 and set @xmath555 .
. then lemma [ lem : poperator ] implies @xmath296 defines a @xmath59$]-linear operator on @xmath74 such that @xmath557 , @xmath558 and @xmath540 .
note that the subspace @xmath559 of @xmath74 is a @xmath5-submodule of @xmath74 .
therefore we define @xmath560 so that @xmath561 is a @xmath5-equivariant projection onto @xmath559 .
let @xmath562 be a short exact sequence in @xmath519 .
since @xmath563 , we have @xmath564 .
therefore @xmath565 is surjective and the left exact functor @xmath533 is actually exact .
similarly , since @xmath566 is flat over @xmath567 , @xmath568 is also an exact functor .
if @xmath569 is the left ideal of @xmath452 generated by @xmath570 , then @xmath571 \rtimes w ) = \dd(v ) \rtimes w / ( \pa_1 , \ds , \pa_n)$ ] and @xmath572 \rtimes w\ ] ] as a @xmath59 \rtimes w$]-module . since @xmath573 \rtimes w ) = k[v^{(1 ) } ] \rtimes w$ ] and the functor @xmath574 is exact , for each @xmath575 \rtimes w\text{-}{\mathsf{mod}}}$ ] we get the standard diagram @xmath576 \rtimes w^k \ar[r ] \ar[d ] & k[v^{(1 ) } ] \rtimes w^l \ar[r]\ar[d ] & n \ar[r ] \ar[d ] & 0 \\ dr \circ dr^{\perp } ( k[v^{(1 ) } ] \rtimes w)^k \ar[r ] & dr \circ dr^{\perp}(k[v^{(1 ) } ] \rtimes w)^l \ar[r ] & dr \circ dr^{\perp}(n ) \ar[r ] & 0 } \ ] ] where the first two vertical morphisms are isomorphisms .
this implies that the third vertical morphism is also an isomorphism .
hence the natural transformation @xmath577 , coming from the fact that @xmath568 is left adjoint to @xmath533 , is an isomorphism .
now take @xmath556 and consider the natural morphism @xmath578 . since @xmath533 is conservative , the fact that @xmath579 implies that this morphism is injective . on the other hand ,
if we define @xmath580 , where @xmath581 , @xmath582 then lemma [ lem : poperator ] implies that @xmath583 .
this proves that @xmath578 is surjective .
it is straight - forward to see that @xmath533 and @xmath568 send graded modules to graded modules .
[ lem : tensorreg ] let @xmath530 be a @xmath59 \rtimes w$]-module such that @xmath585 is isomorphic to @xmath492 copies of the regular representation as a @xmath5-module . then @xmath530 is isomorphic to the regular representation as a @xmath5-module . as a @xmath5-module , @xmath586 .
therefore it suffices to show that the brauer character @xmath587 of @xmath588 satisfies @xmath589 for all @xmath590 in @xmath5 .
recall that @xmath591 / \langle k[(v^*)^{(1)}]_+ \rangle$ ] .
let @xmath592 be the eigenvalues of @xmath590 on @xmath9 . since we are calculating the brauer character of @xmath588 , we assume that @xmath593 for all @xmath173 . then @xmath594 since @xmath595 and @xmath7 does not divide @xmath56 , the product on the right hand side is non - zero . in this section
let @xmath6 be an arbitrary algebraically closed field .
we let @xmath596 denoted a @xmath124-graded @xmath597$]-algebra such that @xmath596 is a finite , free @xmath597$]-module , where @xmath597 $ ] is graded with @xmath598 .
write @xmath599 $ ] and @xmath600 .
for @xmath601 , we denote by @xmath602 the specialization @xmath603 } k_{\alpha}$ ] of @xmath596 at @xmath604 .
we assume that there exists a finite set @xmath605 and @xmath596-modules @xmath606 such that 1 .
the module @xmath607 is graded and free as a @xmath597$]-module .
2 . for all @xmath608
, there is a bijection @xmath609 such that the simple , graded @xmath602-module @xmath610 is the unique simple quotient of @xmath611 , the specialization of @xmath607 at @xmath604 .
3 . there is a bijection @xmath612 such that the simple , graded @xmath613-module @xmath614 is the unique simple quotient of @xmath615 .
let @xmath74 be a @xmath596-module .
we say that @xmath74 is a _ @xmath596-lattice _ if it is free , of finite rank , as a @xmath597$]-module .
note that if @xmath74 is a @xmath596-lattice and @xmath530 a @xmath596-submodule , then @xmath530 is a @xmath596-lattice because @xmath597 $ ] is a principal ideal domain .
fix a graded composition series @xmath621 such that @xmath622 for some @xmath620 .
write @xmath623 for the natural map .
it is an inclusion .
we set @xmath624 then @xmath625 is a @xmath596-submodule of the @xmath596-lattice @xmath607 , hence it is a @xmath596-lattice .
we have a @xmath596-morphism @xmath626 .
we claim that @xmath618 is a @xmath596-lattice .
it suffice to show that it is torsion - free with respect to @xmath597 $ ] .
let @xmath627 and @xmath628 $ ] such that @xmath629 .
then @xmath630 , which implies that @xmath631 .
hence @xmath632 .
since @xmath618 is a @xmath596-lattice and @xmath633 is a simple @xmath613-module , @xmath634 will induce an isomorphism @xmath619 provided @xmath635 .
let @xmath636 , then there exists some @xmath637 $ ] such that @xmath638 and @xmath639 .
therefore @xmath640 . by lemma [ lem : compseries ]
, for each @xmath641 we have a graded @xmath596-lattice @xmath642 such that @xmath643 .
the specialization of @xmath162 to @xmath644 is denoted @xmath645 to distinguish it from the simple module @xmath610 .
let @xmath648 .
since @xmath649 , the set @xmath650 is non - empty . since @xmath596 is noetherian , we can choose @xmath651 to be maximal with respect to inclusion .
the claim is that @xmath652 is prime .
assume otherwise , so that there exist ideals @xmath653 with @xmath654 but @xmath655 .
note that @xmath656 has gk - dimension one and is a finitely generated @xmath657-module .
therefore @xmath657 has gk - dimension one too . since @xmath658 has gk - dimension zero , the short exact sequence @xmath659 implies , by ( * ? ? ?
* proposition 8.3.11 ) , that @xmath660 . however , @xmath661 is clearly a torsion @xmath657-module
. therefore ( * ? ? ?
* corollary 8.3.6 ) implies that the gk - dimension of @xmath661 is zero .
this contradiction implies that @xmath652 is prime .
let @xmath662 .
the @xmath596-submodule of @xmath74 consisting of elements that are torsion with respect to @xmath597 $ ] is a proper submodule of gk - dimension zero . therefore quotienting out by this submodule we may assume that @xmath74 is torsion free and hence free , and that @xmath663
moreover , @xmath664 is a non - zero quotient of @xmath615 which implies that @xmath665 .
this implies that @xmath666 .
since @xmath162 is a @xmath596-submodule of @xmath614 , we have @xmath647 as required . 1
. there exists a finite set @xmath668 such that @xmath669 .
2 . the rank of @xmath162 equals @xmath670 .
the specialization of @xmath162 at @xmath604 is isomorphic to the simple @xmath602-module @xmath610 for all @xmath671 .
let @xmath652 be a prime ideal in @xmath596 such that @xmath647 .
its existence is guaranteed by lemma [ lem : primeann ] .
set @xmath672 so that @xmath162 is an @xmath673-module .
since @xmath673 is a prime ring , the image of @xmath597 $ ] in @xmath673 is a domain .
therefore the fact that @xmath673 is a finite @xmath597$]-module of gk - dimension one implies that @xmath673 is a free @xmath597$]-module . the map @xmath674 $ ] is finite .
let @xmath675 denote the azumaya locus of @xmath673 .
since @xmath675 is open and dense in the irreducible variety @xmath676 and the map @xmath460 is finite , the set @xmath677 is a proper closed subset of @xmath678 . therefore the set @xmath679 is open and dense in @xmath678 .
since the specialization @xmath645 is a quotient of @xmath680 , the simple module @xmath610 is a quotient of @xmath645 .
this implies that every @xmath610 is an @xmath673-module and hence @xmath670 is bounded above by the p.i .
degree of @xmath673 . by definition ,
this bounded is achieved for all @xmath681 .
fix some @xmath681 .
recall that @xmath682 . hence @xmath683 }
l(\lambda ) \ge \dim_k l_{\alpha}(\lambda).\ ] ] on the other hand , since @xmath610 is supported on the azumaya locus of @xmath673 and @xmath614 is a simple module for the central localization @xmath684 of @xmath673 , ( * ? ? ?
* posner s theorem 13.6.5 ) together with ( * ? ? ?
* kaplansky s theorem 13.3.8 ) imply that @xmath685 } l(\lambda).\ ] ] therefore we get the required equality .
we return to the case where @xmath6 has positive characteristic .
recall that @xmath41 is the space of parameters for @xmath144 .
we fix a parameter @xmath37 .
let @xmath686 be a variable and denote by @xmath687 the rational cherednik algebra over @xmath688 $ ] such that the specialization of @xmath687 at @xmath689 recovers @xmath12 .
let @xmath690 be the field of fractions of @xmath688 $ ] .
given @xmath691 , let @xmath692 \rtimes w } \lambda$ ] be the corresponding baby verma module .
it is a free @xmath688$]-module of finite rank .
let @xmath693 \rtimes w } k \o_k \lambda$ ] be the corresponding baby verma module for @xmath694 . 1 .
every simple module @xmath695 is absolutely irreducible .
2 . the module @xmath696 has a unique simple quotient @xmath614 .
we have a natural identification of @xmath694-modules @xmath697 } \bar{\delta}_{\vc}(\lambda ) = \bar{\delta}_k(\lambda)$ ] . 1 .
the @xmath6-algebra @xmath698 is already split semi - simple since @xmath6 is assumed to be algebraically closed and the characteristic of @xmath6 does not divide the order of @xmath56 . therefore the @xmath68-algebra @xmath699 is also split semi - simple .
2 . let @xmath700 , considered as a @xmath701 \rtimes w$]-module such that @xmath701_+$ ] acts as zero .
adjunction implies that @xmath702 \rtimes w}(\bar{\delta}_k(\lambda),\mu ) = \operatorname{\mathrm{hom}}_{w}(k \lambda,\mu ) = k \delta_{\lambda,\mu}\ ] ] since @xmath703 and @xmath704 are absolutely irreducible . therefore ,
if @xmath705 is any simple quotient of @xmath696 , we must have a @xmath701 \rtimes w$]-surjection @xmath706 .
hence @xmath707 is the unique graded quotient of @xmath696 .
3 . the space @xmath708 } \bar{\delta}_{\vc}(\lambda)$ ] is a @xmath701 \rtimes w$ ] such that @xmath701_+$ ] acts as zero .
therefore we have a surjective map @xmath709 } \bar{\delta}_{\vc}(\lambda)$ ] ( since @xmath710 is a generating set for @xmath697 } \bar{\delta}_{\vc}(\lambda)$ ] ) .
since both modules have the same dimension over @xmath68 , this surjection must be an isomorphism .
let @xmath162 be the @xmath687-module described in [ sec : genericl ] . since it is free as a @xmath688$]-module , the graded @xmath5-module structure of all specializations of @xmath162 are the same .
therefore it suffices to prove that the poincar polynomial of the specialization @xmath718 has the desired form .
we fix a filtration @xmath719 of @xmath720 as in lemma [ lem : compseries ] . specializing to @xmath721
gives a filtration @xmath722 of @xmath723 such that @xmath724 .
all the @xmath452-modules @xmath722 have @xmath7-curvature zero .
therefore proposition [ prop : cartier ] says that @xmath725 for some graded @xmath59 \rtimes w$]-module @xmath726 and we have @xmath727 this implies that @xmath728 as graded @xmath5-modules .
let @xmath729 denote the category of @xmath730-lattices i.e. the category of finitely generated , graded @xmath730-modules that are free over @xmath688 $ ] .
it is an exact , extension closed subcategory of @xmath731 .
therefore it makes sense to consider the grothendieck group @xmath732 of @xmath729 . by our assumption on @xmath733 , the support of @xmath716 as a @xmath14-module
is contained in the azumaya locus of @xmath12 .
therefore mller s theorem , ( * ? ? ?
* proposition 2.7 ) , implies that each graded composition factor of the indecomposable module @xmath734 is of the form @xmath735 $ ] for some integer @xmath736 .
hence lemma [ lem : compseries ] says that each @xmath737 specializes at @xmath689 to some @xmath735 $ ] .
thus , in @xmath732 we have an equality @xmath738 = h(t ) [ l_{\vc}(\lambda)]$ ] for some laurent polynomial @xmath739 .
this implies that @xmath740 = h(t ) [ l(\lambda)^{(1)}_{\vc = 0}]$ ] in the grothendieck group of graded @xmath59 \rtimes w$]-modules . by lemma [ lem : tensorreg ]
, the fact that @xmath718 is isomorphic to @xmath492 copies of the regular representation implies that @xmath741 is isomorphic to a graded copy of the regular representation . on the other hand , lemma [ lem : pcochar ] and proposition [ prop : cartier ] imply that @xmath742^{\mathrm{co } w } \o \lambda$ ] , hence @xmath743^{\mathrm{co } w } \o \lambda ] = h(t ) [ l(\lambda)^{(1)}_{\vc = 0}].\ ] ] the proof of ( * ? ? ?
* lemma 3.3 ) implies that @xmath744 . from this one
can deduce formula ( [ eq : poincare ] ) .
assume that @xmath5 and @xmath37 are chosen so that the centre of @xmath12 is smooth .
then every simple @xmath12-module has dimension @xmath745 .
this implies that the graded character of @xmath716 is given by the formula of lemma [ lem : chardiv ] .
since @xmath716 is finite dimensional , the rational function on the right hand side of ( [ eq : poincare ] ) must be a laurent polynomial .
this means that @xmath746 divides @xmath747^{co w},t^p ) i(t)$ ] in @xmath748 $ ] .
the formula ( [ eq : definego ] ) for @xmath749 shows that every root of @xmath605 is a primitive @xmath750 root of unity .
therefore , if @xmath746 and @xmath749 have a non - trivial common factor , we must have @xmath751 for some primitive @xmath750 root of unity @xmath465 .
but @xmath752 .
hence @xmath746 must divide @xmath747^{co w},t^p)$ ] in @xmath753 $ ] .
it was shown in @xcite that if @xmath5 is not isomorphic to @xmath18 or @xmath44 then one can always find some @xmath155 for which @xmath746 does not divide @xmath747^{co w},t^p)$ ] . since @xmath37 has played no part in this argument , we conclude that the centre of @xmath12 is never smooth in these cases .
conversely , it follows from corollary [ cor : gmonensmooth ] that the centre of @xmath40 is regular for generic @xmath37 .
now we consider the group @xmath44 ; we have @xmath754 in this case . by corollary [ cor : equivsmooth ]
, it suffices to show that the blocks of @xmath34 are singletons for generic parameters @xmath37 and all parabolic subgroups @xmath293 of @xmath44 .
we begin by showing that this is true for @xmath755 .
recall from [ sec : defncherednik ] the central element @xmath141 of @xmath756 .
for the remainder of the proof we follow the notation of @xcite . the function @xmath37 is defined by @xmath757 and @xmath758 for some @xmath759 . to show that the block partition of @xmath760 is trivial , it suffices to show that the scalars by which the element @xmath141 acts the @xmath162 s are pairwise distinct .
let @xmath761 and @xmath762
. the scalar @xmath763 by which the central element @xmath206 acts on the simple @xmath44-module @xmath704 is given by @xmath764 we have @xmath765 and @xmath766 .
define @xmath767 .
then @xmath141 acts on @xmath768 by @xmath769 , hence it will act by the same scalar on @xmath768 and @xmath770 if and only if @xmath771 since @xmath772 and the list of values @xmath773 is finite , we can always choose @xmath774 such that ( [ eq : someeq ] ) does not hold , provided there is no pair @xmath775 such that @xmath776 and @xmath777 .
this can be checked directly : e.g. @xmath778 if and only if @xmath779 ; @xmath780 if and only if @xmath781 which implies that @xmath782 and hence @xmath783 - both clearly contradictions . up to conjugacy , there is only one proper parabolic subgroup of @xmath44 , it is @xmath784 .
we may assume that @xmath785 .
repeating the above argument for @xmath784 and noting that @xmath786 , one sees that the block partition of @xmath787 will not consist of singletons precisely if equation ( [ eq : someeq ] ) is satisfied for @xmath788 .
note that , since the p.i .
degree of @xmath12 is @xmath145 , @xmath492 is also the largest power of @xmath7 dividing the dimension of any simple @xmath12-module . also , one can show that @xmath789 is a proper subset of @xmath41 .
first , we show that the subset @xmath790 of @xmath41 consisting of all @xmath37 such that the dimension of every simple @xmath16-module is divisible by @xmath492 is open and dense in @xmath41 .
consider the @xmath791$]-algebra @xmath730 .
it is free as a @xmath791$]-module .
therefore it is a continuous family of @xmath792-algebras in the sense of ( * ? ? ?
* definition 2.2 ) ( for @xmath792 trivial ) .
then ( * ? ? ?
* lemma 2.3 ) says that , for each integer @xmath33 , the set of points @xmath109 such that @xmath16 contains a two - sided ideal of dimension @xmath33 is closed in @xmath41 .
the annihilator of a simple @xmath16-module @xmath74 has codimension @xmath793 ( recall that @xmath6 is assumed to be algebraically closed ) .
therefore the set of all points in @xmath41 for which there exists an ideal in @xmath16 whose codimension belongs to @xmath794 is closed and its compliment is @xmath790 .
we just need to show that this set is non - empty .
take any line @xmath678 in @xmath41
. then lemma [ lem : genericl ] implies that @xmath795 is dense in @xmath678 .
note that this actually shows that there is no linear subspace of @xmath41 contained in @xmath796 .
now we treat the general case .
let @xmath293 be a parabolic subgroup of @xmath5 and @xmath797 the space of parameters for @xmath798 .
restriction of parameters defines a linear map @xmath799 , its image is a non - zero linear subspace of @xmath797 .
we have shown that the set of points in @xmath797 for which all simple @xmath800-module have `` maximum @xmath7-dimension '' is open and dense .
if @xmath801 denotes the pre - image of this set under @xmath802 then it is open and dense in @xmath41 .
then it follows from propositions [ prop : completesmall ] and [ prop : assocgraded ] together with corollary [ cor : equivsmooth ] that @xmath803 where the intersection is over all conjugacy classes of parabolic subgroups of @xmath5 .
we remark that our reduction method ( combining propositions [ prop : completesmall ] and [ prop : assocgraded ] together with corollary [ cor : equivsmooth ] ) also gives a different proof of the result ( * ? ? ?
* corollary 4.2 ) : if @xmath74 is a simple @xmath12 and @xmath318 such that the support of @xmath74 equals the image of @xmath285 in @xmath804 then @xmath805 divides @xmath806 .
when @xmath807 , @xmath808 .
one can show this as follows : by remark [ rem : symsing ] , the algebra @xmath809 is azumaya for all @xmath810 and hence @xmath811 . on the other hand ,
direct calculations show that @xmath812 .
therefore propositions [ prop : completesmall ] implies that @xmath813 . finally , when @xmath49 , @xmath814 and , even though this is not an azumaya algebra , it follows from lemma [ prop : classifysimples ] that @xmath815 does divide @xmath816 for all simple @xmath817-modules @xmath72 . | in this article we study rational cherednik algebras at @xmath0 in positive characteristic .
we study a finite dimensional quotient of the rational cherednik algebra called the restricted rational cherednik algebra .
when the corresponding pseudo - reflection group belongs to the infinite series @xmath1 , we describe explicitly the block decomposition of the restricted algebra .
we also classify all pseudo - reflection groups for which the centre of the corresponding rational cherednik algebra is regular for generic values of the deformation parameter . _ dedicated , with admiration and thanks , to ken brown _ _ and toby stafford on their @xmath2th birthdays _ |
starting from a mechanics problem specified by lagrangian , @xmath43 , the usual mechanical prescription is to require stationary action .
@xmath44 in this report , we will freely substitute force , @xmath45 , and momentum , @xmath46 .
the requirement for stationary action then reads @xmath47 . when @xmath48 , the above procedure directly gives newtonian mechanics and has the advantage of being generally valid under coordinate transformations , @xmath49 .
however eq . [ eq : dadx ] gives second - order equations of motion , requiring @xmath50 by definition , and complicating discussions of numerical integration .
a first - order form can be derived from an alternate lagrangian , @xmath51 by treating @xmath52 and @xmath53 as separate dynamic variables .
using @xmath54 in ( [ eq : dadx ] ) , we find @xmath55 setting these two equal to zero gives a result entirely equivalent to newtonian mechanics , but in which we may consider a verlet - type integration process of updating @xmath56 with fixed @xmath57 , and then updating @xmath57 with @xmath56 fixed .
it has been found that eq .
[ eq : dadx ] forms a solid basis for forming generalizations of physical laws .
a stochastic generalization may be to permit small deviations by constraining @xmath58 in ( [ eq : p ] ) .
the resulting matrices of constraint values ( lagrange multipliers @xmath59 ) can restrict deviations in the action in a history - dependent and non - local way .
each deviation in the action can be thought of as arising from elastic collisions with un - modeled molecules from the surrounding ` bath ' environment or as an unknown lagrangian applied to the system between times @xmath60 and @xmath27@xcite ( with @xmath61 a small time increment ) . without any other constraints ,
this squared - deviation constraint implies interaction with a completely chaotic ( infinite temperature ) bath and does not conserve energy , momentum , etc . except in the deterministic limit ( @xmath62 ) .
when two systems are coupled , the combined system should obey a set of conservation laws .
as formalized by noether s theorem,@xcite such conservation laws can be derived for a single system directly from the action formulation by considering continuous transformations of the trajectory @xmath63 in a region around @xmath64 where @xmath65 .
because the action is stationary with respect to small perturbations in @xmath66 , there exists a vanishing quantity , @xmath67 ( with @xmath68 reserved for denoting transposition ) and it is possible to define an invariant ( using @xmath69 ) .
@xmath70 feynman showed that if the action functional is invariant to this transformation ( @xmath71 = \mathcal a[x(t)+q(t , a)]$ ] ) then the corresponding invariant is a conventional conserved quantity e.g. @xmath72 ( @xmath73 , the ones - vector ) generates the momentum , @xmath74 ( @xmath75 ) generates the energy , etc .
it is instructive to consider a case where the action is not invariant to the transformation .
for example , the form of the action is changed using the substitution for momentum in the simple harmonic oscillator ( @xmath76 ) . however ,
if we assume the existence of a generalized momentum , @xmath77 that is nonetheless an invariant of some ` complete ' system , we can define ( using [ eq : da ] ) @xmath78 as a momentum _
exchange_. if the net momentum of the observed system changes by more than its internal force , it implies that an external system has lost exactly this amount of momentum . in the presence of noise ,
this quantity may not be conserved ( @xmath79 is stochastic ) so that one effect of the noise is to re - distribute @xmath77 over the system and the bath . in general , we can define a set of constraints on @xmath80 using an @xmath81 matrix @xmath82 , whose columns correspond to the constraint directions @xmath83 and whose leftmost column is reserved for the energy @xmath84 . in this case , the vector of exchanges for a given trajectory over a given time interval , @xmath61 , is @xmath85 in this equation , the presence of @xmath61 is used to implicitly denote the stratonovich integral ( see appendix ) .
these considerations have shown a simple method for including the influence of an external system on the dynamics .
if we assume the existence of some `` total '' invariant between the system and the bath , then it makes sense to enforce a stochastic constraint on the average change @xmath86 .
we should note our fixed sign convention , where @xmath87 is always taken to be a quantity belonging to the central system under consideration . before moving on to discuss the obvious connection of these changes to the thermodynamic work
, we shall first consider the significance of the lagrange multipliers in this formalism .
collecting constraints on @xmath88 , @xmath89 and carrying out the maximum entropy procedure specified in eq .
[ eq : rsmax ] for determining @xmath90 for each time , @xmath27 , given a history @xmath91 for @xmath92 , @xmath93\\ & \quad \times z[\beta , x(t')_{t'=0}^{t-\epsilon}]^{-1 } \end{split } \notag \\
\begin{split } \label{eq : trans } & = \exp\left[-({\ensuremath{\tfrac{\delta\mathcal a}{\delta x({t})}}}\epsilon-\mu(t))^t \frac{g}{\epsilon } ( { \ensuremath{\tfrac{\delta\mathcal a}{\delta x({t})}}}\epsilon-\mu(t ) ) + \mu(t)^t \frac{g}{\epsilon } \mu(t ) \right ] \\ & \quad \times z[\beta , x(t')_{t'=0}^{t-\epsilon}]^{-1 } \end{split } \\ c & \equiv ( 2g)^{-1 } \notag \\
\mu(t ) & \equiv c \left ( y(t)\beta\epsilon/2 - \epsilon \int_0^{t-\epsilon } g_{t-\tau } { \ensuremath{\tfrac{\delta\mathcal a}{\delta x({\tau } ) } } } d\tau \right ) \notag .\end{aligned}\ ] ] in this equation , @xmath94 is a vector with the dimension of the system coordinates , since @xmath95^t$ ] .
the action deviation , @xmath96 , then follows a normal distribution with mean @xmath97 and single - time variance / covariance matrix @xmath98 .
the history integral in the above equation could alternatively have been written in terms of a time - dependent covariance function . in this work , however , we will not be concerned with the calculation of history - dependent partition functionals for which this transformation becomes useful . enforcing a constraint for the hamiltonian energy change @xmath99 using the lagrange multiplier @xmath100 ( along with a possibly empty set of additional constraints in the form of eq .
[ eq : di ] ) , leads directly to a generalized langevin equation @xmath101 here the centered ( stratonovich ) wiener process increment has been substituted for the standard normal random variate , @xmath102 , at time @xmath27 using @xmath103 .
it is well - known from the fokker - planck equation@xcite @xmath104 , \ ] ] that the solution to this equation ( in the memory - free case , but see also ref.@xcite ) is the canonical distribution with temperature @xmath105 ( where @xmath106 is the boltzmann constant ) .
this solution is independent of @xmath107 , suggesting a natural parametrization for the langevin equation is in terms of the temperature and @xmath108 , related to the thermal conductivity or rate of temperature equilibration ( see eq . [ eq : temp ] ) .
generally , if an invariant , @xmath87 , can be expressed as a function of @xmath109 , then a result similar to the above should hold for other common equilibrium thermodynamic ensembles as well , such as the @xmath110 ensemble where @xmath111 constitutes an additional dynamic variable and constraint.@xcite this establishes the physical interpretation of the lagrange multipliers as the thermostatic variables of the bath that dictate the eventual equilibrium of the system .
note that increasing @xmath112 tends to decrease @xmath113 , for example increasing pressure will drive the volume downward .
the present work shares some conceptual similarity to the second entropy of attard @xcite . in this report , however , we have been able to derive our results in a mathematically rigorous way directly from two extremum principles , a maximum entropy expression for the transition probability ( eq . [ eq : p ] ) , and constraints derived from an action functional ( eq . [ eq : dadx ] ) .
this allows for trivial generalizations to systems coupled with arbitrary reservoirs .
in addition , there is a clear physical motivation for the transition entropy and the full nonequilibrium entropy production which allows us to find the work done on the system by each constraint as will be shown in the next section .
the above set of equations is also sufficient for defining nonequilibrium analogues of intensive thermodynamic variables such as the temperature .
this can be done by adding a hypothetical constraint , @xmath114 , defined for some set of atoms or region of space in the system .
analogous to the operation of a thermometer ( zero energy exchange on imposing stochastic and damping terms ) to define temperature , we then require that no work is done on average , @xmath115 . integrating using the stratonovich rules developed in the appendix , we find@xmath116 .\end{aligned}\ ] ]
the result is the intuitive kinetic temperature , and is especially simple if we choose @xmath117 as is common for the langevin equation . in that case
, the ensemble average kinetic energy at each instant determines the temperature . for boltzmann - distributed @xmath118 , regardless of the choice of @xmath119 , the average heat flow is zero when @xmath120 determines the temperature .
if different types of particles can be coupled to separate thermostats , such as in plasmas , then it becomes physically meaningful to speak of separate ionic and electronic temperatures . to end this section
, we show that it is possible to derive a brownian limit from our action functional approach using the alternative lagrangian , eqns .
[ eq : l1]-[eq : dadv ] .
first , it can be shown that changes in the hamiltonian are recovered by applying eq .
[ eq : di ] @xmath121\end{aligned}\ ] ] next , applying eq .
[ eq : trans ] and assuming @xmath122 exactly , we find a combined equation @xmath123 the second equation has the form of a generalized brownian motion equation , but includes terms related to the process @xmath53 . in particular ,
if the process @xmath53 becomes unknown , then the best guess form @xmath124 , and a streaming velocity @xmath125 generate an appealing equation for brownian motion . more rigorously ,
if the process @xmath53 is assumed to be unknown , updates @xmath126 should be made based on a stochastic realization of @xmath53 whose average will generate the streaming velocity @xmath127 . to the best of the author
s knowledge , this is a novel derivation of the brownian limit that does not require an explicit limiting process of a infinitely massive particle , or infinite momentum jumps between position changes . instead
, these two assumptions are implicitly present in assuming that @xmath128 are known during each position update .
the memory term derived here is similar to the form postulated in ref .
@xcite , which lead to a quantitative treatment of memory effects .
here we can see it to be a natural consequence of placing constraints on the squared deviation of the action and the energy change at each time - step .
having firmly established the connection to the equilibrium distribution above , we can construct a high - level view of any process employing a series of transition probabilities to effect a change in the state of the system .
this construction will lead naturally to a view of the process in terms of a thermodynamic path transforming one type of energy into another with a concomitant irreversible entropy production . to begin
, we exactly define a system state , @xmath8 , as any information which is known about a system that is sufficient to construct a probability distribution for its variables , @xmath129 .
the machinery of statistical mechanics can then be used to propagate this information to system states at other time - points and under alternate possible processes .
the work of joule and thomson showed that there exists a series of mechanical operations that can be performed to effect a transition between any two thermostatic states,@xcite however this transition can only take place in the direction of increasing entropy .
the entropy increase comes about because of experimental inability to control the detailed motions of all particles , and is therefore zero in the case of completely controllable mechanical work .
thus , it is important to define a mechanical , adiabatic process , in which all work is completely controlled by letting @xmath130 with constant external force experienced by the system @xmath131 .
this is formally a zero - temperature , continuous - time limit , since at at a finite temperature there is some amount of uncertainty about the exact state of @xmath87 on short timescales , which leads to a discrepancy between the force exerted by the external system and its `` long - timescale '' counterpart experienced by the system . in general , only a subset of work values can be controlled , and before proceeding it will be necessary to solidify the concept of controllable work .
if this work is to be delivered by an external thermostatic system , for example an adiabatically coupled piston , then the first law of thermostatics gives @xmath132 if the force can be assumed constant over a sufficiently short time - step .
mechanically , this force corresponds to the force on a wall exerted by a spring placed externally to it .
the total force on the wall is , of course , @xmath133 which implies that if a two spring system were disconnected after a change @xmath134^t$ ] , their internal potential energies would have changed by an amount @xmath135^t \notag .\end{aligned}\ ] ] the sum of these two energy changes is not necessarily zero due to the possibility of momentum change . using the known energy change of the system ,
it should then be possible to solve for the change in kinetic energy of the constraint .
for the system , the _ total _ energy change is given by @xmath136 where we have used eq .
[ eq : da ] for @xmath137 , multiplied by @xmath138 , to define work values and a corresponding `` non - mechanical '' energy transfer , @xmath139 . in the adiabatic limit , all energy transferred to the system by external forces should be reflected by known mechanical changes ( related to @xmath140 ) which is precisely what @xmath141 allows us to do .
note also that unless specifically denoted ` ext ' all quantities refer to the central system , so that @xmath142 means the work done on the system . in the mechanical limit , @xmath143 , the external force experienced by the system in the absence of the thermostatting random noise and @xmath84 .
substituting this quantity from the langevin equation ( [ eq : langevin ] ) , an adiabatic , mechanical system must satisfy @xmath144 the last section gave some physical insight into the quantities @xmath141 .
a mathematical consideration of the previous equation shows that @xmath145 ( @xmath146 being identical to @xmath82 with the first column removed ) can be understood as a projection , removing components of @xmath118 parallel to @xmath147 . if we therefore define ( writing the moore - penrose pseudoinverse of @xmath8 as @xmath148 ) @xmath149 then @xmath150 identically .
for an example application , @xmath73 ( ones - vector ) is associated with the system momentum , and eq .
[ eq : dota ] generates the average velocity @xmath151 when @xmath152 .
[ eq : dota ] is invariant to multiplication of @xmath119 by a constant , and so persists in the deterministic limit .
the work done on the system ( eq . [ eq : dw ] ) is @xmath153 note the information - theoretic quality of the work defined by the above equation .
if separate reservoirs existed that were able to independently influence the motion of each particle in the system , then @xmath146 would become an identity matrix , and @xmath154 would equal @xmath155 identically . in the presence of noise , the work done does not necessarily equal the energy change . in the @xmath73 example above
, the work is @xmath156 here `` @xmath157 '' denotes a sum over @xmath158 ( one - dimensional ) atoms , with obvious extension to multiple dimensions . because this interaction controls only the total system momentum , the work is computed using the average velocity change . as shown in the appendix ,
the expectation of this stochastic integral is @xmath159 .
similarly , an electric field can couple only to the net dipole moment of a system .
this implies the transformation of applied energy to heat if it can not be manipulated to perfectly match fluctuations in the driven variable .
the kinetic energy change of @xmath87 , ascribed to the reservoir , can be determined in the mechanical limit from @xmath160 using eqns .
[ eq : due ] and [ eq : dw ] . ] .
the above equations thus completely describe any exchange of mechanical energy between deterministic systems exerting known forces .
if the two - spring system considered above were disconnected at time @xmath161 , an outside observer absorbing the kinetic energy of the wall ( eq . [ eq : dk ] ) , the energy change of the reservoir would reduce to the usual thermostatic potential change @xmath162 .
a well known consequence of liouville s theorem is that the entropy change is zero in a completely deterministic process.@xcite using adiabatic processes , then , it is possible to propagate a starting state , @xmath8 , to any state with constant entropy . if , however , phase space volume were not preserved , it would amount to a discarding of information on the state of the system at a given time ( for example , by integrating the probability over short time intervals ) .
then the amount of work that can be recovered from the system will become less than the amount input . in an extreme case ,
all information about the system may have been lost , flushing the corresponding information content to zero . starting from this unknown state , @xmath163 ,
the probability of a given frequency distribution , @xmath164 is approached by @xmath165}$ ] , with @xmath166 $ ] the familiar information entropy functional @xmath167 = -\int p_i \ln p_i / p^0_i .\ ] ] further , if exchanges of conserved quantities during some process @xmath168 are known , then the set @xmath169 are also known from @xmath8 , and this information can be usefully employed to increase the amount of work that can be recovered showing the entropy as a measure of `` lost work '' .
it therefore stands to reason that any adiabatic process should be described by not only the above mechanical work values , but also the change in information entropy due to information loss .
next , consider allowing heat exchange in addition to controllable work .
assuming an infinitely large reservoir ( or a short enough time - step ) , added heat will cause a negligible change in reservoir temperature . because we have assumed the work done on each reservoir ( [ eq : due ] ) can be reversibly stored , these are not associated with an entropy change .
we therefore introduce the physical entropy change in the reservoir as due only to exchange of heat , or `` non - work '' energy , @xmath170 , with @xmath171 , and @xmath172 originating from heat removed from the system plus a kinetic energy , @xmath173 , assumed to be recoverable only as heat . in this work ,
different notations are used for the information entropy , @xmath38 , the physical entropy , @xmath161 , the caliber , @xmath174 , and the caliber - like functional , @xmath175 .
all of these quantities are defined to be unitless and have some relation to the information entropy of eq .
[ eq : s ] .
@xmath176 the energy rejected to the reservoir as heat can alternately be understood as the total energy dumped to the environment minus the energy removed `` reversibly '' ( @xmath177 ) .
this interpretation shows that if some of the changes in the environment were re - classified as irreversibly stored , so that the information @xmath178 becomes lost , then this is equivalent to adding that energy to the total @xmath172 .
the total energy rejected to the environment is recoverable ( the mechanical limit considered above ) if and only if @xmath179 , implying @xmath180 .
these considerations again highlight the subjective nature inherent in the definition of irreversibility .
connecting back to the usual thermostatics , @xmath181 plays the role of an energy for the reservoir .
@xmath182 where we must provide an experimental justification for the ability to use or store the energy terms appearing in the above sum . for any transformation , @xmath168 , the total entropy change deriving from information loss and interactions with the environment
is given by the change in information entropy plus the heat exchange term @xmath183-{\ensuremath{\mathcal h}}[p(x|a ) ] + \beta_0 q_\text{ext } .\ ] ] it should be noted that this formula is still not complete if there is a change in the phase space between @xmath8 and @xmath9 , for example if particles are added / removed , or if the state space is uniformly dilated . in this case , we have extra information on the region of phase space occupied after a transition .
in general , if the state at time @xmath12 is known to be @xmath184 , then the size of the region of configuration space accessible at time @xmath13 is @xmath185 ( eq . [ eq : z ] ) .
this reduces the entropy at @xmath13 to @xmath186 $ ] . writing this down for each transition , @xmath187}{{\ensuremath{\mathcal{p}_{i}\left({x_i}\right)}}z_i[-\beta , x_{i+1 } ] } - di^t \beta .\end{aligned}\ ] ] the above equation has been symmetrized by including the corresponding entropy decrease for @xmath12 given that @xmath184 was inferred from @xmath188 , using the forward step probability but with reversed forces , @xmath189 .
support for this form is given in the next section , where it is also shown that @xmath190 using the gibbs inequality .
nevertheless , eq . [ eq : ent1 ] can already be applied to langevin and brownian motion with uniform diffusion constants . as another example , applying eq .
[ eq : ent1 ] to a process taking any point to the equilibrium distribution shows that the expected entropy change for this process is the usual system entropy difference plus @xmath191 . because the end - point entropies are fixed , alternate processes for transforming @xmath168 are restricted to varying @xmath192 . for such processes ,
we may write a strong form for the clausius form of the second law @xmath193 here , @xmath194 is a function of the end - points , and @xmath120 and @xmath172 are fully variable along the path . choosing a path from a fully specified distribution , @xmath8 , to a maximum entropy distribution , @xmath195 , and then to the ending maximum entropy distribution @xmath169 , we may employ a quasistatic , `` reversible , '' path between the two maximum entropy distributions , so that @xmath196 .
the heat evolved in this best - case process has its origin in the re - classification of information that occurs during the coarsening of @xmath8 , in accordance with conclusions on maxwell s demon.@xcite the equilibrium theory is therefore contained in the present development in the form of slow , quickly relaxing processes .
this perspective shows the intimate connection between coarse - graining that assumes infinitely fast relaxation of the reservoir and the traditional theory of quasi - static processes .
however , the complete theory also permits an investigation of both relaxation processes and entropy production in time and history - dependent processes moved arbitrarily far from equilibrium by coupling to simple thermostatic reservoir systems .
as noted in the introduction , there is a fundamental difference between applying maximum entropy to a complete trajectory and to each transition probability distribution individually . referring to eqns .
[ eq : p ] and [ eq : z ] , the probability for a path @xmath197 ( using a starting distribution @xmath198 on @xmath199 ) is @xmath200}}{\mathcal z[\{\lambda\},\{x\}_0^{s-1 } ] } { \ensuremath{\mathcal{p}\left({x_0}\right ) } } \\
\eta[\gamma;\{\lambda\ } ] & \equiv \sum_{i=0}^{s-1 } \sum_k \lambda_{k , i } f_k(x_{i+1},\{x\}_0^i ) \\ \mathcal z[\{\lambda\},\{x\}_0^{s-1 } ] & \equiv \prod_{i=0}^{s-1 } z[\lambda_i,\{x\}_0^i ] \\ { \ensuremath{\mathcal{p}_{0}\left({\gamma|x_0}\right ) } } & \equiv \prod_{i=0}^{s-1 } { \ensuremath{\mathcal{p}_{0}\left({x_{i+1}|\{x\}_0^i}\right ) } } .\end{aligned}\ ] ] defining a path information entropy , or caliber , associated with the number of possible ways that a given @xmath201 could be observed , we have @xmath202 this equation shows that the path entropy is an average of maximum entropy increments given by eq .
[ eq : rs ] . this form telescopes in the following way .
suppose only the process from time @xmath203 to @xmath204 is of interest . in this case , the above entropy expression can be separated from @xmath205}\right\rangle } } } + \sigma_{\gamma ' } \\ \intertext{with the non - anticipating partial sum } \sigma_{\gamma ' } & = \sum_{i=0}^{j-1 } { \ensuremath{{\left\langle{\sum_k \lambda_{k , i } f_k(x_{i+1},\{x\}_0^i)}\right\rangle } } } + { \ensuremath{{\left\langle{\ln z[\lambda_i,\{x\}_0^i]}\right\rangle } } } + s_0 .\end{aligned}\ ] ] for times between @xmath206 , the first line above has exactly the same form as eq .
[ eq : caliber ] , with @xmath207 replaced by @xmath208 .
that is , the starting distribution has become a multi - step probability distribution with no change in the total caliber .
analogous to the equilibrium theory , we should expect that a cumulant expansion of an appropriate path free energy will yield path averages , fluctuations , etc . from eq .
[ eq : caliber ] , a path free energy functional can be defined as @xmath209 \equiv -{\ensuremath{{\left\langle{\ln\mathcal z}\right\rangle } } } = { \ensuremath{{\left\langle{\eta}\right\rangle } } } - \sigma_\gamma + { \ensuremath{\mathcal h}}_0 .\ ] ] expanding @xmath210\right ) { \ensuremath{\mathcal{p}\left({\gamma}\right ) } } \ ; d\gamma \\ & = { \ensuremath{{\left\langle{f_k(x_{i+1},\{x\}_0^i)}\right\rangle } } } - \int \ln \mathcal z { \ensuremath{\frac{\partial { } } { \partial { \lambda_{i , k } } } } } { \ensuremath{\mathcal{p}\left({\gamma}\right ) } } \ ; d\gamma \\ & = { \ensuremath{{\left\langle{f_k(x_{i+1},\{x\}_0^i)}\right\rangle } } } + { \ensuremath{{\left\langle{f_k(x_{i+1},\{x\}_0^i ) - { \ensuremath{{\left\langle{f_k(x_{i+1},\{x\}_0^i)}\right\rangle}}}}\right\rangle } } } \\ & = { \ensuremath{{\left\langle{f_k(x_{i+1},\{x\}_0^i)}\right\rangle } } } .\end{aligned}\ ] ] we see that the derivatives of @xmath211 indeed give the path averages , @xmath212 .
this result is valid for any chosen set of @xmath20 and corresponding functions , @xmath213 .
it is therefore possible to formally use the above to compute arbitrary path expectations , even if they do not affect the dynamics ( @xmath214 ) .
we can also use the above equation to formulate a first law of time - dependent , nonequilibrium thermodynamics @xmath215 = \sum_{i=0}^{s-1 } \sum_k { \ensuremath{{\left\langle{f_k(x_{i+1},\{x\}_0^i)|\lambda}\right\rangle } } } d\lambda_{k , i } .\ ] ] we can similarly compute second derivatives to give @xmath216 .\end{aligned}\ ] ] using the legendre transformations ( @xmath217 ) given in eq.ns [ eq : f]-[eq : df ] , these fluctuations can be transformed to ensembles with constrained averages ( acceleration , particle flux , etc . ) , @xmath212 , rather than thermodynamic forces , @xmath20 .
a simple derivation for such equations can be given following ref .
@xcite . performing a second - order expansion using eq .
[ eq : gibbs ] , and writing the result in matrix form , @xmath218 we swap the sides of a set the averages @xmath219 and their corresponding forces @xmath220 . @xmath221 re - assembling the matrices on each side , and inverting @xmath222 , @xmath223 the above manipulation , well - known in the theory of linear , local equilibrium processes,@xcite are seen as gibbs relations in the general nonequilibrium theory developed here .
these equations describe a change of ensemble in the equilibrium theory @xcite . in this respect , they form a basis for connecting stochastic , langevin dynamics simulations ( e.g. eq . [ eq : langevin ] ) to constant kinetic energy , solute flux , etc .
ensembles studied extensively in nonequilibrium molecular dynamics simulations@xcite .
in addition , the upper - right matrix element , @xmath224 { \operatorname{cov}[f_2,f_2]}^{-1 } d{\ensuremath{{\left\langle{f}\right\rangle}}}_2 $ ] appears as the starting point for the mori projector - operator method@xcite as has been discussed by jaynes.@xcite using the action - type conserved quantity constraints ( eq . [ eq : di ] ) in the free energy functional , its derivatives generate the fluxes @xmath225 . at equilibrium , these fluxes are zero . given information about the previous history of the system , @xmath226 , we can use eq .
[ eq : gibbs ] to find the linear change @xmath227 this is the linear equation of motion for a near - equilibrium system under a mechanical driving force , and explains the ubiquitous use of fourier transforms in solving these equations .
it could be expanded to arbitrary order using a taylor series in @xmath228 .
the case of driving by thermal , or constant external force , conditions is much simpler , and we may accordingly treat the more complicated case of a transient process . according to eqns .
[ eq : one ] and [ eq : gibbs ] , we may expand about an arbitrary initial distribution plus some reference program , @xmath229 ( denoted by the zero subscript ) , to find the equation of motion for thermal driving . @xmath230
this relation contains a factor of @xmath231 , as in a derivation by searles and evans @xcite where it was shown to reduce to the green - kubo expression in the zero - field limit .
as noted in that work , the last term on its own is incorrect when a driving force is present .
the leading term in the above is the flux in the reference process . choosing this reference process as a conducting steady - state shows one mechanism for the failure of the onsager reciprocity relations .
the relations @xmath232 should hold in any case , but are only derivatives of the flux / force curve . these derivatives are analogous to the fluctuation moments of the canonical ensemble , which can approximate the energy at a slightly altered temperature .
they are expansions about a fully nonlinear free energy functional , @xmath211 .
the linear relations derived in the last section should not be expected to hold for large deviations in the nonequilibrium forces .
we can progress beyond this limitation by analogy to the transition from thermodynamic integration to free energy perturbation in equilibrium free energy calculations .
any two processes on the same path space , @xmath233 , can be compared using a likelihood , @xmath234 } \equiv \frac{{\ensuremath{\mathcal{p}_{b}\left({\gamma}\right)}}}{{\ensuremath{\mathcal{p}_{a}\left({\gamma}\right ) } } } = e^{-(\eta_b-\eta_a ) } \frac{\mathcal z_a[\gamma]}{\mathcal z_b[\gamma ] } .\ ] ] the likelihood obeys @xmath235}}\right\rangle}}}_a = 1 $ ] , and ( by the gibbs inequality ) @xmath236}\right\rangle}}}_a \le 0 \le { \ensuremath{{\left\langle{l_{a\to b}}\right\rangle}}}_b$ ] .
the distribution of the weight satisfies the perturbation formula@xcite @xmath237}\right ) } } & = \int e^{l_{a\to b}[\gamma ] } \delta(l - l[\gamma ] ) { \ensuremath{\mathcal{p}_{a}\left({\gamma}\right ) } } \ ; d\gamma \\ & = \int \delta(l - l[\gamma ] ) { \ensuremath{\mathcal{p}_{b}\left({\gamma}\right ) } } \ ; d\gamma \\ & = { \ensuremath{\mathcal{p}_{b}\left({l = l_{a\to b}[\gamma]}\right ) } } .\end{aligned}\ ] ] it is possible to express path averages using the above quantity as @xmath238}\right\rangle}}}_b = { \ensuremath{{\left\langle{a[\gamma]e^{l_{a\to b}[\gamma]}}\right\rangle}}}_a .\ ] ] however , in the equilibrium theory , a constant related to the free energy difference is usually cancelled on the right - hand side of this expression .
there is no such constant in the above equation because we have not identified an appropriate extensive variable . in the present case
, we can use the free energy ( eq .
[ eq : pathf ] ) to define @xmath239 } & \equiv \frac{{\ensuremath{\mathcal{p}_{b}\left({\gamma}\right ) } } e^{-\mathcal f_b[\lambda ] } } { { \ensuremath{\mathcal{p}_{a}\left({\gamma}\right ) } } e^{-\mathcal f_a[\lambda ] } } \\ \intertext{in terms of which } { \ensuremath{{\left\langle{e^{l'_{a\to b}[\gamma]}}\right\rangle}}}_a & = e^{-(\mathcal f_b[\lambda]-\mathcal f_a[\lambda])}\\ \intertext{and } { \ensuremath{{\left\langle{x[\gamma]}\right\rangle}}}_b & = \frac{{\ensuremath{{\left\langle{x[\gamma]e^{l'_{a\to b}[\gamma]}}\right\rangle}}}_a}{{\ensuremath{{\left\langle{e^{l'_{a\to b}[\gamma]}}\right\rangle}}}_a } .\end{aligned}\ ] ] or course , the above expressions may be of little analytical use because @xmath240}\right\rangle}}}$ ] has simply been subtracted from @xmath241 $ ] ( in the exponent of @xmath242}$ ] .
however , the transition probabilities usually adopt particularly simple forms , e.g. normal distributions .
if , for example , the transition probability takes every point to an equilibrium distribution , then the above expansion is particularly useful .
it has been claimed that entropy production can be gauged by the ratio of forward to reverse path probabilities.@xcite for the process derived in sec .
[ sec : dyn ] , we can define a reverse process by inverting the sign of the generalized forces , @xmath243 , and normalizing the distribution separately for each @xmath188 .
such a reversal corresponds to an attempt at guessing whether energy has been added or subtracted during a step @xmath16 , with the action deviation constraint , @xmath107 , unchanged .
thus , we can define a ratio @xmath244 } { { \ensuremath{\mathcal{p}\left({x_{i+1}}\right ) } } p_0(x_i|x_{i+1 } ) z [ \beta , x_i ] } e^{-di^t \beta } \notag .\end{aligned}\ ] ] since @xmath245 by bayes theorem , this result exactly matches eq .
[ eq : ent ] arrived at through thermodynamic reasoning .
moreover , several steps can be concatenated to give @xmath246 the above equation can therefore be viewed as a statistical basis for the second law of thermodynamics .
it is physically motivated by observing that the entropy increase is attributed to a combination of environmental entropy changes , @xmath172 , and information - like entropy changes , @xmath247/p_0(x_{i+1 } ) } { { \ensuremath{\mathcal{p}\left({x_i}\right ) } } z[-\beta , x_{i+1}]/p_0(x_{i } ) } \\\ ] ] the above term implies an extra contribution to the total entropy change beyond eq .
[ eq : ent1 ] . to understand this contribution ,
consider a simple one - state system , to be transformed into a two - state system through the ( unbiased ) transition probability @xmath248 .
the presence of the normalization in eq .
[ eq : ent ] leads to an additional factor of @xmath249/p_0(x_{i+1 } ) = -\ln 2 $ ] in the entropy .
however , this term is canceled by the entropy of the resulting state , @xmath250 , so that the total entropy change for this process is zero .
if , instead , we perform a bit - set operation by going in the opposite direction , the entropy will increase if @xmath251 is any non - uniform state corresponding to our loss of information on discarding the bit .
a term of this form is exactly what we should have expected when writing down eq .
[ eq : ent1 ] . applying this equation to a situation where a particle is added to the system , we find that the re - normalization will physically compensate for the expansion of phase space which leads to a difference in entropy measures .
mathematically , this implies that the `` default '' measure @xmath252 is replaced by the re - normalized measure @xmath253 if we have information about the previous state when calculating the entropy at state @xmath13 . equation [ eq : two ] is connected to the likelihood definition ( [ eq : l ] ) . defining @xmath9 as a `` forward '' process , starting from @xmath198 and employing @xmath254 , and @xmath8 as a `` reverse '' process , employing @xmath255 . because @xmath256 and the end - point distributions @xmath257 have been pre - determined , we can write eq .
[ eq : ent ] as @xmath258 this completes the connection between the physical entropy production defined in sec . [ sec : thermo ] and fluctuation theorems of the form ( [ eq : l ] ) .
however , the entropy production has not yet been connected to the caliber functional ( eq . [ eq : caliber ] ) .
to do this , we decompose eq .
[ eq : ft ] into the form of eq .
[ eq : caliber ] as @xmath259}\right\rangle}}}_b + { \ensuremath{\mathcal h}}_s .\end{aligned}\ ] ] although this definition is similar to the caliber , it is not an information entropy , since the averages are taken with respect to the forward probability distribution , @xmath260 .
the choice of the forward direction corresponds to the direction in which information propagates @xcite .
it determines the target distribution for taking the divergence ( eq .
[ eq : divs ] ) . in the special case where a stationary distribution is known during each time - propagation step , a set of useful equalities can be derived simply from the re - weighting equation @xmath261 the first average represents a single - time average over the steady state distribution used to propagate the system during the transition @xmath262 .
the second average is a path average over stochastic trajectories beginning in the steady - state at time zero .
the proof@xcite is by recursion from the property of the stationary distribution under time - propagation , @xmath263 these relations will work for any stationary distribution of the transition probability .
equality of the starting and ending temperatures is not required .
in order to give concrete examples of eqns . [ eq : gibbs ] and eq . [ eq : ent ] , we must choose a set of coarse variables of interest ( @xmath212 ) and follow their time - evolution , @xmath264 .
most mesoscopic models contain hydrodynamic equations of motion for the solution density .
a rigorous route@xcite to their derivation is by forming suitable integrals of the probability distribution function appearing inside the liouville equation , and much of the early literature on nonequilibrium problems is focused on this derivation . in the maximum transition
entropy context , the resulting equations describe the propagation of a state of knowledge forward in time using an exact equation of motion .
the exact internal and external forces on the system are required using this route , and the projector - operator formalism is used to add the uncertainty introduced by mixing processes occurring below the size and time - scale of the density function . in the stochastic formalism developed here , the liouville equation has been replaced with the klein - kramers and smoluchowski equations .
in addition to convection , these equations also describe diffusion of probability that occurs because of loss of information during mixing processes . summing the fokker - planck equation for eq .
[ eq : brown ] over particles of each species type , @xmath265 , and integrating over coordinates other than that of a single , distinguished molecule ( using the relations from ref .
@xcite ) gives the one - particle evolution equation @xmath266 in this process , the force and diffusion coefficients have become averages over the probability distributions of the molecules of each type i.e. a potential of mean force . here , @xmath267 has been substituted for @xmath268 .
the form of equation [ eq : diffn ] corresponds exactly to the usual diffusion equation used for the continuum formulation of electrodiffusion equations @xcite , without the necessity of considering the damping or particle radius and stokes - einstein relation . in the one - particle approximation
, there is an information loss associated with integrating over the full n - particle distribution , @xmath269 . because less information is used to propagate the system , there is a start - up entropy production of @xmath270-{\ensuremath{\mathcal h}}[\rho(x , p , t ) ] \ge 0 $ ] that is due to a loss of useful work that may be extracted from the system without this information .
time - evolution followed by discarding information on particle correlations then transforms eq .
[ eq : ent ] into @xmath271-{\ensuremath{\mathcal h}}_\text{max}[\rho(t ) ] - di_c^t\beta \ge 0 .\ ] ] for the fokker - planck equation of brownian motion ( [ eq : diffn ] ) , the only information used to propagate the distribution is the density at each time - step , and the maximum entropy distribution is the momentum - free distribution with known average spatial density and average energy.@xcite in the special case where the energy is a local function of the density , this leads to the well - known local equilibrium theory@xcite .
the momentum and non - diffusive conservation equations may also be derived from the full langevin equation ( [ eq : langevin ] ) in a manner similar to that shown in ref .
@xcite . although more complex , these hydrodynamic systems are amenable to the analysis given above
. the entropy production will be a combination of heat production and information loss .
if the system moves between two steady - states , the entropy production will be given simply by the end - point information entropy difference plus the integral of the evolved heat divided by the temperature of the external heat reservoir .
these quantities are both linked to information loss because the microscopic details of the energy exchange have been replaced by less informative probability distributions over the states of both systems . for a numerical calculation
, we may turn to the one - dimensional example of an optically trapped bead or an atomic - force microscope pulling experiment .
assuming very fast relaxation of the momentum of the pulling coordinate over a potential energy surface , @xmath272 , it is appropriate to use eq .
[ eq : brown ] with the average velocity set at zero.in the case of an optical trap with center @xmath273 and force @xmath274 , we can employ a change of coordinates to @xmath275 ( with @xmath276 ) so that during each time - step , the position and trap center are updated to give @xmath277 the constraint on @xmath278 , determines the rate at which the bead is allowed to dissipate energy into the solution , while the energy constraint , @xmath112 , acts as a driving force for net energy exchange . in the brownian case , the system can not distinguish between internal and external forces , so that eq .
[ eq : dw ] shows energy decrease as work from all applied ( assumed internal ) forces is dissipated into the surroundings as heat at each time - step .
reversing our sign convention to treat the harmonic trap as external , the work done on the bath through the system is @xmath279 would also have been converted into heat from the bath system s internal potential , @xmath272 . , showing that this term can be characterized as a coupling to an external thermostatic system . ]
the entropy increase of eq .
[ eq : ent ] ( eq .
[ eq : ft ] ) is formally a path average , and its evaluation requires specifying an initial state and a driving protocol . as in ref .
@xcite , we may choose to follow several velocity programs starting from a steady - state at constant pulling force .
@xmath280 the ` housekeeping heat ' dissipated by the bath s removal of the bead s momentum at each time - step leads to a steady - state dissipation of @xmath281 . after a sufficiently long time , the instantaneous information entropy will reach its steady - state value , @xmath282 = ( 1-\ln\tfrac{\beta\kappa}{2\pi})/2 $ ] .
note that the entropy of the steady - state distribution is well - defined because it is invariant to the change of coordinates @xmath283 . for constant pulling force and temperature
, this expression says that over long time - periods the integral of the information entropy change will be zero , and the total entropy increase will be due completely to dissipated work . during intermediate time - periods ,
however , the entropy increase will be a combination of changes in the position distribution plus the dissipated work .
@xmath284 because the dynamics is markovian , the average dissipated work can be easily calculated from the distribution at each time - step and the total of each type of work will be a sum of one - step stochastic integrals ( eq .
[ eq : dwf ] ) . assuming the distribution is gaussian at a starting time and fourier - transforming eq .
[ eq : smof ] gives a gaussian distribution at all future times for all driving protocols , whose mean ( @xmath285 ) and variance ( @xmath286 ) are solutions of first - order ordinary differential equations .
@xmath287 using @xmath288 , we can easily verify @xmath289 \ge 0 .\ ] ] using numerical integration , we have plotted ( fig .
[ fig : drive ] ) the rate of entropy production vs. time for two hypothetical driving protocols .
the units in the figure are the same as in ref .
@xcite , and their third protocol ( followed by its time - reversal starting at 0.28 s ) has been used for the upper two sets of panels . because the variance of the distribution only responds to changes in diffusion constant , temperature or driving force ,
we have varied @xmath290 in the second set of calculations .
the information entropy rate gain goes to zero and the heat production becomes constant at the onset of the eventual steady - state .
as shown by the heat production during compression from @xmath291 to 5 pn/@xmath285 m , excess heat production is required whenever the information entropy decreases .
when the distribution expands , heat production decreases as the mean begins to lag behind the trap center and the distribution expands .
this is counter - balanced by an increase in information entropy , leading to net dissipation .
, solid line ; force constant , @xmath290 , long dashed line on the left scale ; and response of the mean , @xmath285 , on the right _ vs _ time ( s ) .
the right set of panels show the decomposition of the entropy increments ( eq . [ eq : incr ] , pn-@xmath285 m ) into heat ( solid ) and information gain / loss ( long - dashed ) .
whenever the information entropy decreases , an equal or larger amount of heat is produced so that the total ( eq .
[ eq : totsf ] , short - dashed ) is always positive . for the upper two sets of panels , the force constant was held constant at @xmath292 pn/@xmath285 m so no change in information entropy occurs .
for the lower two , the distribution is compressed , then broadened by changing @xmath290 between 3 and 5 using a cubic interpolation lasting 80 ms.,width=319 ] we note that a large amount of additional complexity can be added to this model by adding information about the variables here treated as ` external ' to the description of the dynamics .
if local variations in the fluid velocity or temperature were included , then the dynamics would have to specify the equations of motion for these fields .
the final entropy increase may then be more or less than this result because these degrees of freedom may be responsible for additional heat production , but more information on the fluid state has been included , leading to decreased information loss .
in this paper , we have given a generalization of the theory for driven , irreversible processes .
a set of transition probabilities defines the evolution equation for the system of interest .
a special simplification is the case of langevin and brownian motion , which can be recovered as limits of a constrained action integral approach .
non - anticipating stochastic trajectories for classical particle and field motion can be cast in this form .
the action functional interpretation gives a physical method for defining conserved quantities and the energy cost associated with transfers of these quantities from an external environment or experimental apparatus .
deterministic dynamics is recovered from the langevin equation when the deviation of the action functional is strongly constrained to zero ( sec .
[ sec : dyn ] ) . in this limit ,
the external forces which appeared as statistical in the stochastic approach become mechanical . because both limits appear in this derivation , the fluctuation - dissipation theorems derived as gibbs relations from eq .
[ eq : one ] are applicable in the case of both thermal and mechanical driving forces .
these equations are completely general in the sense that they apply not only arbitrarily far from equilibrium , but also during transient processes which do not possess a steady state .
a particularly useful aspect of this approach is that it directly connects multiple length and time - scales .
the formulation of the equations has been in terms of particle motion , but coarse - grained relations are easy to define as appropriate ensemble averages over these motions .
examples of such averages include centers of mass for polymer units or average density and velocity fields .
the coarse equations of motion will then lead to polymer coarse - graining models@xcite or non - local hydrodynamic models.@xcite for the time - evolution of average quantities , we expect the thermodynamic limit argument@xcite to apply when the number of averaged degrees of freedom is large so that the path realized by the system under a given set of constraints will fall arbitrarily close to the maximum entropy solution an overwhelming majority of the time .
the present work is therefore a suitable foundation for the theory and analysis of nonequilibrium molecular dynamics .
applications to simplified , standard examples such as circuit theory are easily accomplished .
the joule heating of a resistor , for example , can be seen from eq .
[ eq : pdiss ] as fundamentally arising from the difference between the velocity added to each ion individually _ vs. _ the usable energy in the average ion velocity . because the energy added to the system in driving the ions
is not expressible in terms of the average velocity alone , spreading in the distribution of ion velocities becomes heat .
the same remarks follow for driven convective transport , where a spreading in the distribution of forward fluid momentum leads to increases in the local temperature ( eq . [ eq : temp ] ) .
connections of this theory to the formal structure of maximum entropy thermodynamics and bayesian inference have been elaborated upon in ref.@xcite these connections allow the definition of thermodynamic cycles expressing differences between driving protocols using the same free energy techniques commonly employed in the equilibrium theory . some examples have already appeared in the literature for path re - weighting@xcite .
it is expected that expression in terms of thermodynamic cycles will greatly simplify the derivation and interpretation of these studies .
we have identified a new generalization of the second law for irreversible processes .
a traditional analysis shows that the total entropy increase ( eq . [ eq : ent ] ) is dependent on the details of system dynamics and exchange of conserved quantities with an external system . connecting this with the fluctuation theorem ( eq . [ eq : ft ] ) gives a microscopic form for the second law of thermodynamics .
the physical device of tracking work performed on individual particles as well as external reversible work sources allows us to track the flow of each type of work ( and heat ) through the system .
because these changes come directly from the forces on each degree of freedom , this analysis does not depend arbitrary decompositions of energy functions or definitions of steady - states . from an informational perspective
, entropy increase comes about from discarding information and/or from the information loss associated with coupling to external reservoirs .
this is distinguished from the entropy production functional of local equilibrium theory in that the entropy functionals developed here include long - range correlations and are not necessarily extensive.@xcite it is a simple matter to define more complicated baths , for example affecting only the average temperature in a given area for imposing thermal gradients .
it should be noted that the analysis in sec .
[ sec : thermo ] showed that increasing the number of controllable variables decreases the number of degrees of freedom associated with heat production .
molecular insertion and deletion operations will aid in generalizing this approach to include imposed chemical potential ( insertion force ) as well as particle flux boundary conditions , but have not been considered here .
this work was supported , in part , by sandia s ldrd program .
sandia national laboratories is a multi - program laboratory operated by sandia corporation , a wholly owned subsidiary of lockheed martin corporation , for the u.s .
department of energy s national nuclear security administration under contract de - ac04 - 94al85000 .
despite the wealth of literature on the langevin and wiener processes , the procedure for calculating expectations of time - integrals given in standard references such as gardner@xcite and risken@xcite remains complicated . because stratonovich integrals appear prominently in the present paper , often usurping the role of thermodynamic potentials , we present here two alternative methods .
both rely on replacing expressions to be evaluated at the midpoint of a time - step with the first - order expansion , @xmath293 . using eq .
[ eq : di ] with @xmath294 to find the energy change , we expand the average velocity about the mid - point , @xmath295 \\ & = ( i-\tfrac{\epsilon\beta_0}{4}m^{-1}c)\dot x_i + \tfrac{\epsilon}{2}m^{-1}(f - c\tilde y\tilde\beta/2 ) + \tfrac{1}{2}m^{-1}c^{1/2}{\ensuremath{d\mathcal w}}+ o(\epsilon^{3/2 } ) \\ & = \dot x_i + \tfrac{\epsilon}{2 } m^{-1}\dot p ' + o(\epsilon^{3/2 } ) , \end{aligned}\ ] ] where @xmath296 is computed using only quantities at the time - step @xmath12 . multiplying this with @xmath297 from eq . [ eq : langevin ] , we get @xmath298 since @xmath299 is normally distributed with mean zero and variance - covariance matrix @xmath300 , @xmath301 has a noncentral @xmath302 distribution with expectation @xmath303\ ] ] for a single constraint , @xmath304 , we find a definition of the kinetic temperature , eq . [
eq : temp ] .
if , in addition , we include a constant pulling force , @xmath305 , we find @xmath306 .\ ] ] what emerges is a kinetic temperature with respect to the terminal velocity , @xmath307 , as well as a heating term .
as discussed in the text , these integrals should also result from differentiating a partition function ( eq . [ eq : f ] ) .
we present an extended derivation of the main results of this approach here .
both langevin and brownian equations can be derived as appropriate limits of the constraints @xmath309 where @xmath310 and @xmath311 introduce external forces .
next , we make the one - half step substitutions , @xmath312 using the force - per - mass , @xmath313 , the appropriate derivative matrices @xmath314 , and defining @xmath315 , @xmath316 .
factoring @xmath317 , gives a normal distribution for @xmath318^t\epsilon$ ] with penalty matrix ( inverse of the variance - covariance matrix , keeping terms below @xmath319 ) @xmath320 and mean ( to first order in @xmath61 ) @xmath321 these expressions are in accord with eqns .
[ eq : langevin ] and [ eq : brown ] .
note that the fokker - planck equation can be used to prove that the boltzmann distribution is stationary under either the langevin ( @xmath323 ) or brownian ( @xmath324 ) limits , but not both . for the langevin limit
, it can be checked that the derivative of this equation with respect to @xmath325 gives eq .
[ eq : langen ] . for the brownian limit , we find again eq .
[ eq : dwf ] .
these rely on the following expansion for the derivative of the log - determinant term @xmath326 since @xmath327 should contain a prefactor of @xmath61 , the second term is usually unimportant , so that @xmath328 robert zwanzig , k. s. j. nordholm , and w. c. mitchell .
memory effects in irreversible thermodynamics : corrected derivation of transport equations . ,
5(6):26802682 , jun 1972 .
correction to zwanzig , 1961 .
richard lehoucq and anatole von lilienfeld - toal .
translation of walter noll s derivation of the fundamental equations of continuum thermodynamics from statistical mechanics " .
, 100:524 , 2010 .
10.1007/s10659 - 010 - 9246 - 9 . | we derive a physically motivated theory for non - equilibrium systems from a maximum entropy approach similar in spirit to the equilibrium theory given by gibbs . requiring hamilton s principle of stationary action to be satisfied on average during a trajectory ,
we derive constraints on the _ transition _ probability distribution which lead to a path probability of the onsager - machlup form .
additional constraints derived from energy and momentum conservation laws then introduce heat exchange and external driving forces into the system , with lagrange multipliers related to the temperature and pressure of an external thermostatic system .
the result is a fully time - dependent , non - local description of a nonequilibrium ensemble coupled to reservoirs at arbitrary thermostatic states .
detailed accounting of the energy exchange and the change in information entropy of the central system then provides a description of the entropy production which is not dependant on the specification or existence of a steady - state or on any definition of thermostatic variables for the central system .
these results are connected to the literature by showing a method for path re - weighting , creation of arbitrary fluctuation theorems , and by providing a simple derivation of jarzynski relations referencing a steady - state .
in addition , we identify path free energy and entropy ( caliber ) functionals which generate a first law of nonequilibrium thermodynamics by relating changes in the driving forces to changes in path averages .
analogous to the gibbs relations , the variations in the path averages yield fluctuation - dissipation theorems .
the thermodynamic entropy production can also be stated in terms of the caliber functional , resulting in a simple proof of our microscopic form for the clausius statement .
we find that the maximum entropy route provides a clear derivation of the path free energy functional , path - integral , langevin , brownian , and fokker - planck statements of nonequilibrium processes .
physical considerations justify a fundamental definition of thermodynamic entropy increase as system information entropy plus heat exchange with an external thermostatic system .
the green - kubo fluctuation theorems@xcite relate equilibrium time - correlation functions with the time - response of system observables to an external driving force .
they are important for their ability to calculate the transport coefficients appearing in onsager s phenomenological relation @xcite , @xmath0 which identified @xmath1 with the entropic " driving forces , @xmath2@xcite .
these two relations form a rough draft for the a first law of nonequilibrium thermodynamics .
combined with the usual second law prescription of increasing entropy , the above establishes the direction in which the system will relax toward equilibrium and an estimate of the entropy increase by this process .
however , the linear transport equations have been derived by analogy with the equilibrium theory , and their interpretation must be made with respect to the entropy of a quasi - steady process .
a more satisfactory development would therefore determine the range of conditions for which eq .
[ eq : onsager ] holds , as well as provide a foundation for studying processes with arbitrary driving forces and defined without reference to any equilibrium state . if , in addition , the theory was able to resemble the well - known equilibrium statistical mechanics , it would offer a wealth of immediate insight into new applications for which now standard nonequilibrium methods may prove cumbersome and error prone
. such a resemblance must be reached by defining path functionals analogous to the energy and entropy of equilibrium states and would therefore constitute a true statistical mechanics for thermodynamics ( as opposed to thermostatics )
. our first question in this investigation will be how such a microscopic first law of thermodynamics can be formulated .
although much work has been devoted to the above problems , these questions have been addressed from a large number of different viewpoints in the last thirty years .
further , there appears to be little consensus on a unifying , general , set of relations from which which all nonequilibrium results may be derived.@xcite recent work has centered on a second question of deriving a microscopic second law of thermodynamics through proving the existence of fluctuation theorems@xcite and exploring their consequences for nonequilibrium systems .
a fundamental fluctuation theorem result is an expression for the `` lost work '' over and above the equilibrium free energy for stochastic processes that convert one thermostatic state to another@xcite via mechanical driving .
this lost work can be interpreted as an entropy increase .
it is simple to show in the case of an isolated deterministic system@xcite @xmath3 where the work , @xmath4 , is the gain in system energy @xmath5 , and both distributions are required to be at the same temperature , @xmath6 , to use the equilibrium relation @xmath7.@xcite further work has provided examples of many fluctuation theorems.@xcite each can be used to define a measure of irreversibility , since for any two path probabilities , @xmath8 and @xmath9 , @xmath10 \ge 0 , \end{aligned}\ ] ] where @xmath11 is the kullback - leibler divergence ( and necessarily positive ) .
although special significance is often attributed to time reversal ( defined by replacing the time evolution operator from @xmath12 to @xmath13 with a reversal of odd functions of time at step @xmath13 , unaltered time evolution from @xmath14 to @xmath15 , and another reversal of odd functions at @xmath12),@xcite this operation is not possible if one - way steps are present , so that only one of @xmath16 or @xmath17 has probability zero .
furthermore , the extension of this formalism to systems lacking momentum , such as discrete processes described by a transition probability matrix is also unclear ( although suggestions involving a [ possibly non - unique ] stationary distribution ) have been offered@xcite .
until these issues are resolved , there does not exist a definitive path functional able to answer the first question above .
early works were principally focused on the first question .
after a detailed picture of how time - correlations control the rate of relaxation to equilibrium@xcite , a general set of relations ( projector - operator theory@xcite ) was described from which all such macroscopic relaxations may be derived.the principal content of the theory was to define a coarsening , projection , operator which removes information about the unmodeled degrees of freedom .
the exact , non - markovian kinetic equation for the probability distribution in the space of remaining variables then shows the relaxation process in the form of a time convolution of the time - correlation functions and the thermodynamic forces driving them to equilibrium.@xcite that is , eq . [ eq : onsager ] should be replaced with @xmath18 where @xmath19 . despite conjectured relationships to a nonequilibrium entropy function,@xcite its relation to the thermostatic entropy
has not yet been fully justified in terms of the maximum entropy formalism or clarified to the point where it is possible to derive the above from a second derivative@xcite analogous to the gibbs relations for equilibrium which give rise to obviously related quantities such as heat capacity , compressibility and coefficient of thermal expansion . at this point ,
the situation shared a peculiar similarity to the circumstances surrounding gibb s classical text introducing the principle of maximum entropy@xcite .
the method for deriving fluctuation dissipation theorems ( analogous to the first law ) was to define a system evolving according to an exact lagrangian , make a random phase approximation to yield an ensemble of exactly evolving trajectories , and then derive a corresponding `` physical '' distribution on trajectory space ( analogous to phase - space , tbl .
[ tbl : corresp ] ) .
because of the prevailing attitude regarding mechanics as the only possible method for solving such problems at the time , gibbs use of maximum entropy methods was seen as a non - physical trick@xcite to derive properties of molecular equilibrium . again , introducing maximum entropy lead to an expansive generalization of the fluctuation dissipation theorem by jaynes@xcite who , it should be noted , has recognized and written about the above conflict of gibbs@xcite .
jaynes introduced maximum entropy following the program of gibbs and using the subjective interpretation of probability used by laplace and jeffreys.@xcite making the substitution from a state ( containing a variable at a single time ) to an entire trajectory immediately identifies a non - equilibrium analog of the gibbs ensemble , complete with entropy and free energy functionals on trajectory space . pursuing the analogy further , jaynes showed that the first derivatives of the path free energy yielded averages of path functionals , and the second derivatives ( minus the space and time - correlation functions ) give their `` first - order '' perturbation with respect to changing the thermodynamic forces , @xmath20@xcite
. therefore we can shorten eq .
[ eq : zwanzig ] to @xmath21 for _ any _ path functional , @xmath22 , we wish to average and _ any _ set of `` controlled '' quantities @xmath23 which define our trajectory ensemble .
if , for example , @xmath23 included time and space - dependent particle fluxes ( i.e. @xmath24 indexes both time and space so that @xmath25 ) , then eq . [ eq : jaynes ] easily generalizes eq .
[ eq : onsager ] to fluctuation relations defined directly from the set of constrained path functionals , @xmath26 .
notice that the perturbation expansion above is not defined by reference to an equilibrium state , but instead with respect to a reference probability distribution on path space .
examples of such elegant derivations of fluctuation theorems have been given many times in works employing path integral methods@xcite such as those related to the onsager - machlup action@xcite .
further work@xcite has also shown how jaynes path entropy may be connected to fluctuation and entropy production theorems .
however , eq . [ eq : jaynes ] contains a fatal flaw . to see this
, first note that the average of a quantity at time @xmath27 is dependent on the `` control '' parameters throughout the whole trajectory .
this is because logical inference does not contain a preferential time direction , and knowledge that the system has a given property at time @xmath28 constrains the state at time @xmath27 .
although this could be alleviated by requiring only casual information to enter into the determination of the state at time @xmath27 , this approach leads to a probability distribution valid only for @xmath29 and not any previous times .
a better approach is to maximize the entropy of the transition probability distribution .
limiting the scope of the maximum trajectory entropy procedure in this way automatically corrects a related re - normalization problem .
suppose phase space were to branch at a future time @xmath28 . in this case ,
a uniform measure on path space would assign points at time @xmath27 a different weight depending on future events .
a simple example is the monty hall problem with a prize assigned to the first door without loss of generality.@xcite the contestant s choice of one of three doors plus monty opening another ( not concealing a prize ) constitutes four possible paths , and maximum path entropy would give each path an equal weight an intuitive , but incorrect , solution .
the correct probability assignment is a uniform distribution for each _ transition _ , leading to ending weights of @xmath30 for the two paths following the selection of the first door , ( contestant s correct choice plus monty choosing randomly from the remaining doors ) , and @xmath31 ( contestant s two incorrect choices plus monty s forced choice of an incorrect door ) . ] . in more abstract terms
, the marginal probability at each time should not depend on the future a concept expressed mathematically by defining a progressively measurable function with respect to the natural filtration of stochastic processes@xcite . in practice
, this flaw can be avoided by considering only processes where assigning equal _ a priori _ weight to all paths is equivalent to assigning equal weight to all transitions . by liouville
s theorem , this is obviously true for deterministic processes .
more generally , the approaches are equivalent when the number of possible transitions does not depend on the starting point .
the path entropy approach of jaynes is therefore valid in the absence of factors re - normalizing for starting - point dependent differences in the number of possible paths ( @xmath32 = z[\lambda_i]$ ] in eq .
[ eq : z ] ) .
a further question remains on the application of fluctuation - dissipation theorems to both langevin and brownian ( overdamped langevin ) processes .
what is the appropriate order for coarse - grained equations of motion ? writing down a langevin equation assumes newtonian , second - order dynamics , whereas the first - order brownian motion can also be derived via the same approach .
although the green - kubo relations were supposed to have solved these problems , ambiguity remains from this approach at a fundamental level because a strict derivation using the method of mechanics does not offer direct insight into the choice of macroscopic variables used for systematic coarse - graining .
in fluid mechanics , either can be applied , and the choice between full newtonian motion models or simplified advection - diffusion models is predictably based on the scales of length , relaxation time , and applied force involved .
several authors have made substantial contributions@xcite in connecting fluid mechanics and non - equilibrium dynamics . although seemingly incompatible , there are advantages to both the mechanics - based and the statistical derivations .
for example , considering the choice of equations of motion from the statistical perspective of equation [ eq : jaynes ] at once allows us to see the consequences of each choice .
if we choose a first - order equation of motion , then we substitute the velocity at a point for @xmath22 and expand it about a streaming velocity @xmath33 in the particle fluxes , @xmath34 .
if second order , then @xmath22 becomes a change in momentum at a point and we expand about the average force , @xmath33 , in the stress tensor of the surrounding fluid , @xmath35.we believe that further corrections sometimes employed in fluid mechanics may also be derived via extending the process that lead to eq . [
eq : jaynes ] .
if both the nonequilibrium first and second laws could be addressed from the same perspective , it seems that an expansive generalization of non - equilibrium statistical mechanics could be achieved . in order to combine these two viewpoints ,
we introduce the following maximum entropy argument .
suppose the information about the time change of a set of dynamic variables , @xmath36 , consists in a set of prescribed averages , @xmath37 .
using only this information , we are to construct a probability distribution for the dynamic variables at point @xmath13 , given a known @xmath36 at time @xmath12 .
according to the standard maximum information entropy ( @xmath38 ) machinery , the answer is @xmath39 - 1 ) -
\sum_{k=1}^m \lambda_{k , i } f_k(x_{i+1},x_i ) \big ] \end{split } \\ \label{eq : p } p(x_{i+1}|x_i ) & = p^0(x_{i+1}|x_i ) e^{-\eta_i[x ] } / z[\lambda_i , x_i ] \\ \label{eq : eta } \eta_i[x ] & \equiv \sum_{k=1}^m \lambda_{k , i } f_k[x_{i+1},x_i ] \\ \label{eq : z } z[\lambda , x_i ] & = \sum_{\{x_{i+1}\ } } p^0(x_{i+1}|x_i ) e^{-\eta_i[x ] } \\ \label{eq : rs } { \ensuremath{\mathcal h}}_{i+1|i } & = \ln z[\lambda , x_i ] + { \ensuremath{{\left\langle{\eta_i|x_i}\right\rangle } } } \\ \label{eq : drs } d{\ensuremath{\mathcal h}}_{i+1|i } & = \sum_{k=1}^m \lambda_k d{\ensuremath{{\left\langle{f_k|x_i}\right\rangle } } } \\ \label{eq : f } \psi_{i+1|i } & \equiv \sum_{k=1}^{m ' } \lambda_k { \ensuremath{{\left\langle{f_k|x_i}\right\rangle } } } - { \ensuremath{\mathcal h}}_{i+1|i } \\ \label{eq : df } d\psi_{i+1|i } & = \sum_{k=1}^{m ' } { \ensuremath{{\left\langle{f_k|x_i}\right\rangle } } } d\lambda_j - \sum_{k = m'+1}^m \lambda_k d{\ensuremath{{\left\langle{f_k}\right\rangle } } } . \end{aligned}\ ] ] these are the expressions relating to the statistical state at time @xmath13 given information on the transition probability distribution . the functional notation for quantities such as @xmath40
$ ] has been used to indicate that in general , these may be considered as functionals depending on the trajectory over all times before @xmath13 . in the continuous limit
, the above quantities exist between times @xmath12 and @xmath13 , and should be viewed in the stratonovich definition .
the appendix shows example calculations of the partition function for the wiener process . the essential difference between eq .
[ eq : p ] and the maximum path entropy prescription is that in the latter , the path probability is @xmath41-\beta u(x_0)}}{z } { \ensuremath{\mathcal{p}_{0}\left({\{x_i\}_0^s}\right ) } } , \\ \intertext{for some path functional , $ a$ , and constant , $ z$ , while this is replaced by } { \ensuremath{\mathcal{p}\left({\{x_i\}_0^s}\right ) } } & = \prod_{i=0}^{s-1 } \frac{e^{-\eta_i[x_{i+1};x_i]}}{z_i[x_i ] } { \ensuremath{\mathcal{p}_{0}\left({\{x_i\}_0^s}\right)}}\end{aligned}\ ] ] in the former .
the present conditional expansion implies a dual characterization of a stochastic process as a single ensemble of paths and a set of telescoping ensembles , each with well - defined , non - anticipating energy changes through @xmath42}\right\rangle}}}$ ] .
.correspondence between single - time and time - dependent path maximum entropy formulations of statistical mechanics . [ cols="<,<",options="header " , ] [ tbl : corresp ] using the above , maximum transition entropy , form has several distinct advantages for the derivation of non - equilibrium relations .
not least is the correspondence to the canonical , maximum entropy form of equilibrium thermostatics pointed out in table [ tbl : corresp ] .
we begin by deriving the generalized langevin and brownian dynamics from a consideration of the action deviation as a constrained quantity . in sec .
[ sec : thermo ] , we connect the langevin equations derived from our stochastic action deviation principle to energy exchange and thermodynamic entropy production via interaction with external reservoirs.@xcite following jaynes information theoretic derivation of statistical mechanics , we identify a quantity analogous to a free energy and entropy functionals for maximum entropy transition probability processes in sec .
[ sec : pred ] .
the coefficient of thermal expansion , isothermal compressibility , and heat capacity can be derived using second derivatives of the equilibrium free energy . in an analogous way
, we show how green - kubo transport theory can be derived from second derivatives of legendre transforms of the maximum transition entropy functional .
these relations can describe the response of both steady and non - steady states arbitrarily far from equilibrium .
applications of this result enable calculation of derivatives of the current - voltage curve at constant current or constant voltage and give general conditions under which onsager reciprocity will hold .
we then connect these functionals to the thermodynamic entropy production and crooks - jarzynski fluctuation relations . |
paradoxes hit us with some mystery .
the simpler and the shorter their formulation , the more appealing and the more frequently they are discussed on scientific fora and at meetings .
paradoxes challenge our understanding and often insist on conceptual clarity .
they have appeared about everywhere in scientific discourse , some having a long history .
an ancient example is democritos paradox , as described for example in _ nature and the greeks _ by erwin schrdinger @xcite : take a cone , or make it something more tasty like a pear , and slice it anywhere parallel to its base .
the two circular faces thus produced must have the same area ; they just perfectly fit together .
but , if they are equal in size , how could the pear ever get its cone - shape ?
that paradox touches at the foundations of statistical mechanics because it deals with the relationship between atomism as a physical theory and the continuum nature of macroscopic objects .
it witnesses to the long struggle , as apparent also in zeno s paradox , of reconciling the continuum with discreteness , of describing physical limits in a correct mathematical language when starting from corpuscular concepts of nature .
another famous now 20th century example is the twin paradox originating in special relativity and first described by paul langevin .
the problem with the twin has little or nothing to do with statistical mechanics ( except when one would discuss the influence of traveling on metabolic processes ) and no confusion on atomism or on thermal aspects can arise .
it just teaches us about the essential structure of minkowski space - time and the invariant meaning of proper time .
however , when combined with quantum field theory , the problem touches on the one of accelerating travelers observing black - body radiation at the unruh temperature , having the same form as the hawking temperature of a black hole .
time after time then paradoxes induce discussions on the physical interpretation of the formalism of our best theories or on possible extensions or unifications .
+ sometimes however history repeats itself , and paradoxes are brought forward that have been answered long ago .
it probably means that the paradox is still much alive and deserves further elaboration when formulated in a new context .
it could also suggest that simple things have been forgotten or that sloppy thinking has gone unnoticed . whatever is the case ,
the problems or paradoxes that are discussed below are very present in today s scientific communication .
they appear in introductions of talks ; there are books written about them , and the web has very many entries repeating naive mistakes against some of the foundations of statistical mechanics .
it is likely that there are different and much more interesting and deeper issues associated to the horizon problem and to the information paradox , but the present contribution discusses these paradoxes as encountered most often in the streets of physics . and the suspicion of th .
smiths is then unavoidable : can there not just be some misunderstanding here of basic statistical mechanics ?
+ as the title suggests , the present paper is not truly a specialized scientific one in the more traditional sense .
it is more popularizing and perhaps provocative at times , hoping it can stimulate discussions on fundamental questions in high energy physics from the point of view of statistical mechanics .
a more technical paper concentrating on sharper discussions of the information paradox is under construction ( joint work with wojciech de roeck ) .
the usual statement of the so called _ horizon problem _ is at best naive and at worst fundamentally misconceived .
here is the wikipedia version ( 20 february 2015 ) @xcite : _ the horizon problem is a problem with the standard cosmological model of the big bang which was identified in the late 1960s , primarily by charles misner .
it points out that different regions of the universe have not `` contacted '' each other because of the great distances between them , but nevertheless they have the same temperature and other physical properties .
this should not be possible , given that the transfer of information ( or energy , heat , etc . ) can occur , at most , at the speed of light .
_ the notion of thermodynamic equilibrium has various aspects , and it obviously depends on considered spatio - temporal scales and types of observations . forgetting much of these last subtleties and speaking operationally , equilibrium is the thermodynamic condition of macroscopic systems where there is a homogeneous temperature , chemical potential(s ) and pressure . in that way , an equilibrium system has no systematic currents or collective motions of energy or particles , subsystems are again in equilibrium and their dynamical condition is that of detailed balance , reversibility and not showing any difference between evolution forwards or backwards in time .
many more features can be added , and depending on the physical situation ( e.g. on how the system is open to the environment ) various ensembles can be used to describe equilibria mathematically , each governed by their own thermodynamic potential ( and corresponding variational principle ) and associated gibbs machinery with its thermal probabilities as introduced also by maxwell and boltzmann . the theory and its mathematics
are very well developed , a highlight of 20th century physics , including the fundamental understanding of phase transitions , critical phenomena and possible instabilities related to long range interactions ( such as the jeans instability for gravity ) , @xcite .
+ there is however a deeper statistical understanding of the equilibrium condition , which starts from basic observations on the particularities of systems composed of a very large number of constituents .
to unveil already the key - ingredient : the law of large numbers is to be expected to play a fundamental role in any description in terms of additive quantities involving a massive number of terms .
+ i start from the simplest set - up for the phase space of classical mechanical systems at fixed energy @xmath0 , volume @xmath1 and particle number @xmath2 .
it represents all the allowed microscopic states @xmath3 ( positions and momenta of all the particles in the system ) and indeed we take as working hypothesis that all states @xmath3 on that energy surface are equally probable ( microcanonical distribution ) . that liouville measure is rather natural , unbiased as it is and invariant for the mechanical evolution .
indeed the hamiltonian evolution defines a flow in that space , which is reversible and incompressible .
+ we can try to classify the macroscopic conditions of the system by first defining a number of macroscopic quantities and to see what are the possible values .
the transition from microscopic states @xmath3 to macroscopic condition ( values of macroscopic variables ) is formalized by a many - to - one map @xmath4 . generally @xmath5 is achieved through spatial averaging ( as when computing a density ) or by counting averages ( like the fraction of particles with a given property ) .
that map roughly induces a partition on the constant energy surface , dividing it into patches of all states @xmath3 that have the same macroscopic value @xmath5 .
the largest patch is the condition referred to as equilibrium .
that induces a partition of the phase space in `` rooms '' or phase space regions that distinguish between macroscopic conditions .
+ it goes without saying that the coarse graining is physically inspired and the choice of macroscopic variables is not completely arbitrary : we prefer macroscopic descriptions that are sufficiently simple and yet in a way , are dynamically closed ( e.g. giving rise to first order dynamical evolutions on hydrodynamic scales ) .
here there is a role for the specific dynamics but in fact already earlier the hamiltonian has entered as we are fixing the energy and we only consider microscopic states with that energy .
+ the rest of the story is statistical and is based on another property of the relevant macroscopic quantities : they are arithmetic averages over space or over the various particles of local quantities . for example
, the density in mass or energy can show a macroscopic profile as made from the various local concentrations of masses or energy .
it is here that the law of large numbers starts to play : typically , when randomly selecting a phase space point , it belongs to the phase space region of ` thermal equilibrium , ' where macroscopic quantities take their equilibrium values .
it is by very far the largest room in the phase space , as visualized in the figure above . for many - particle systems
the equilibrium region will be overwhelmingly huge compared to the other ( nonequilibrium ) regions and in that way the equilibrium values are _ typical _ values just as in the law of large numbers .
no details about the system or its dynamics have been specified yet except that we fixed the energy and that we want our macroscopic description to be relevant and ( in a sense ) complete ( for macroscopic autonomy ) .
irrespective of that , it tells us that equilibrium is the most probable condition from the macroscopic point of view .
+ perhaps we can as well remember the * boltzmann entropy * , given by @xmath6 where @xmath7 of a phase region @xmath8 denotes its liouville volume , always given the constraints on @xmath2 , @xmath0 and @xmath1 .
it quantifies the plausibility of a macroscopic condition , in the sense that @xmath9}{\text{prob}[m ' ] } = e^{[s(m ) - s(m')]/k_b}\ ] ] where @xmath10 stand for macroscopic conditions , outlook , values , ... take here into account that the entropy @xmath11 scales with the number of particles , so that an entropy change of order 1 joule / kelvin is easily reached in the kitchen .
such an increase of entropy easily gives rise to a factor @xmath12 for the ratio of probabilities .
+ i repeat that equilibrium ( as sensed for example by homogeneous temperatures ) is in that sense quite the opposite of `` special . '' here
is how the master himself was saying that : + _ one should not forget that the maxwell distribution is .... in no way a special singular distribution which is to be contrasted to infinitely many more non - maxwellian distributions ; rather it is characterized by the fact that by far the largest number of possible velocity distributions have the characteristic properties of the maxwell distribution , and compared to these there are only a relatively small number of possible distributions that deviate significantly from maxwell s . _
( ludwig boltzmann , 1896 ) + wait , do not read on without having understood that citation .
the horizon problem belongs to the original theoretical motivations for inflationary cosmology .
the general feeling there is that , as in the standard cosmological model no causal process can establish thermal relaxation within the presently observable universe , we need inflation ( a period of accelerated expansion ) to account for the _ observed _ thermal homogeneity
. there may well be other reasons to believe in inflation or , what would be even better , solid observational evidence for the process of inflation , but that is not discussed here .
+ here is the usual presentation of the horizon argument , @xcite : there is something special and even very implausible about having equal temperatures in distant not causally connected regions of our universe , as was measured with relative fluctuations of @xmath13 on the wmap for cosmic microwave ( black body ) radiation and confirmed by the planck esa mission . that specialness of equal temperatures can be removed and hence understood by an inflation scenario ( somehow pushing back in time the big bang ) as that would allow thermalization via causal contact to have taken place after all .
we refer to the nasa - page @xcite for a summary of that standard formulation . in the following section
i address the above two claims @xcite .
i start with a summary of the answer .
equal temperatures are the rule in equilibrium and equilibrium is typical .
if indeed temperature is found to be almost homogeneous , then there is no need and indeed no use in explaining that via thermal relaxation unless you know for sure that the temperatures were not equal at a previous time , if defined at all .
in fact , the universe never was and still is not in equilibrium concerning the gravitational degrees of freedom ( including products as ourselves ) , but there is no reason to doubt its large scale homogeneity in temperature at any moment in its evolution . + here is my evaluation of the reasoning based on the above boltzmann picture : no , on the contrary , there is nothing special about equal temperatures .
in fact , equal temperatures are typical for all regions which are solely constrained to conservation of energy .
if we imagine the universe with the standard cosmology according to a friedmann - lematre - robertson - walker geometry @xmath14 with at an initial time @xmath15 shortly after the big bang an arbitrary matter distribution with a given total energy , then we can and should expect uniform temperature all over .
that is just the statement that equilibrium is typical . as a matter of logic
, thermalization makes the universe less special so that thermalization can not explain specialness ; the universe would have needed to be more special before . in other words requiring thermalization is not only not needed ; it is worse than useless .
+ the above comments concerning the horizon problem are not original ; they have been around for some time , and have been written down more or less in the same way by a number of people ; see in particular the analysis by roger penrose @xcite on the horizon problem and personally i have greatly benefitted from discussions with shelly goldstein ; see also @xcite .
+ we can also add that a rough calculation ( in communication with frederik denef ) based entirely on equilibrium fluctuation theory shows that the fluctuations as measured by the wmap are bigger ( not _ smaller _ ) than expected .
the calculation uses that for black body radiation the specific heat scales like @xmath16 . for volume
@xmath1 we take one pixel of the wmap cmb image , which is a volume much bigger than a cube light year .
for @xmath17 we can take a temperature of about 1000 kelvin at the time of cmb emission . the relative standard deviation in energy scales like @xmath18 .
hence we get @xmath19 . +
it should not surprise us however that there is such a deviation as clearly gravitational degrees of freedom are extremely important here .
the universe was indeed very special at the big bang .
as far as we know and can reasonably assume ( also based on the fact that all verified calculations on nuclear synthesis and chemical reactions in the early universe are based on standard equilibrium thermodynamics ) , there appears to have been a thermal equilibrium but not for the gravitational degrees of freedom which prefer a clustered ( macroscopic ) matter distribution .
ever since , the universe is relaxing its gravitational degrees of freedom along the einstein equation of general relativity . at any rate ,
if we want to know why the universe was initially so very special for its macroscopic gravitational / geometric condition , we need completely different arguments from those alluded at in the horizon problem . if we enquire why it was thermally in equilibrium , there is and remains boltzmann s answer : that is only normal ; it is a matter of counting .
+ note that the above is not saying that contact or dynamics would not play a role in the _ approach _ to equilibrium .
derivations of e.g. diffusive behavior are not at all simple . but even there , it is not _
just _ the dynamics that matters and statistical reasoning on typicality of initial conditions will play a crucial role .
maxwell characterized a proposed reduction of the second law of thermodynamics to a theorem in dynamics , with + _ as if any pure dynamical statement would submit to such an indignity _
( letter to tait , 1876 ) .
the truth of the second law is as in a statistical theorem , _ of the nature of a strong probability ... not an absolute certainty _ like dynamical laws .
( foreword in a book by tait , 1878 ) . in the above
i have in boltzmann s picture not addressed specifically the case of long range interactions ( such as in newtonian gravity ) .
the ideas are indeed not different , but the consequences look different than for a dilute gas . as i said , gravity typically leads to clustering and that is not at all in contradiction with boltzmann s ideas , on the contrary .
note here that some problems can be created ( not solved ) by taking the so called canonical ensemble for treating a system of particles with ( very ) long range interactions ; as is well - known the canonical and microcanonical ensemble need not always be equivalent and there are special problems with the physical meaning of the canonical ensemble when the difference between bulk and boundary fades .
the microcanonical treatment for systems with gravity does not present any special conceptual differences with that of dilute gases , but of course equilibrium looks totally different .
we have thus also emphasized that the hot big bang exactly by its supposed matter homogeneity is a very nonequilibrium state of affairs .
yet , an even better understanding of gravity from a statistical mechanical point would certainly be welcome @xcite .
more specifically i have in mind that a good unison between general relativity and ( nonequilibrium ) statistical mechanics has not at all been found .
the book by richard tolman , _ relativity , thermodynamics and cosmology _ , written in 1934 must be revisited @xcite , at last .
in fact only in the last decade or so a renewed interest has been observed in kinetic relativistic gas theory and that is still restricted to special relativity .
so the horizon problem should get us moving , and remains valuable , if not as motivation for an inflation scenario , then to get us to start thinking about questions as : + - what are the kinetic constraints in the geometric relaxation of the universe to equilibrium ?
+ - what are the roles of expansion and gravitational instability in a statistical mechanical description of the universe ?
+ - what are the relevant and correct macroscopic variables in a geometric theory of gravity ? how to quantify here the distance to equilibrium ?
is there an associated boltzmann entropy which satisfies an h - theorem ? + what is the quantum statistical mechanical equilibrium of a gravitational system hawking radiation ?
+ - how can one formulate the balance equations of irreversible thermodynamics in general relativity ?
the so called information paradox consists of multiple questions and problems related to the construction of a quantum theory of gravity .
it turns out that our understanding today is not optimal , and in particular that shows up in various specific attempts that run into inconsistencies .
that is very normal for a scientific domain in full development , but it is not very specific .
in fact it is not easy to get a sharp and precise version of the paradox .
that black hole information paradox appears in many different versions , changing in time and depending on the source @xcite .
what it is generally accepted to imply is that our usual effective descriptions do not seem to work .
it is heard that our effective quantum field theory in which we have good reasons to trust , produces conflicting or inconsistent results .
surely it could very well be that the quantum field theory that is usually applied there is inappropriate , or must be extended ; the claim that follows below is much more conservative : even _ within _ the usual scheme of quantum field theoretical understanding of hawking radiation , the problem is false ; there is no inconsistency for th
. smiths . or , the usual arguments leading to the paradox are not well founded statistical mechanically .
does it mean that we do understand a quantum theory of black holes ?
no , but let us concentrate on the real problems , and the information paradox if formulated in any sharper way does not appear to be one of them . here
is again a wikipedia version ( 20 february 2015 ) @xcite : _ physical information seems to ` disappear ' in a black hole , allowing many physical states to devolve into the same state , breaking unitarity of quantum evolution . from the no - hair theorem
, one would expect the hawking radiation to be completely independent of the material entering the black hole .
nevertheless , if the material entering the black hole were a pure quantum state , the transformation of that state into the mixed state of hawking radiation would destroy information about the original quantum state .
this violates liouville s theorem and presents a physical paradox .
_ these previous lines do return in many of the ( more ) relevant references , but i do not include here a list of the most popular ones .
there is for example the book _ an introduction to black holes , information and the string theory revolution .
the holographic universe _ by leonard susskind and jameson lindesay @xcite , or the collection of papers in _ quantum aspects of black holes _ , ed .
xavier calmet , springer fundamental theories of physics , @xcite .
we do not comment on the way entropy or the second law are presented there , and move directly to the core of the argument ( basically chapters 8 - 9 in @xcite and the paper on the firewall phenomenon by r.b .
mann in @xcite ) .
we also do not discuss the precise meaning of the firewall proposal and how its introduction was thought to avoid some `` entanglement paradoxes , '' @xcite .
+ to find a somewhat precise formulation of the information paradox is not so easy for th .
there are many web - entries , some very instructive as prepared e.g. by samir mathur @xcite .
a very clear version is in @xcite , also having the advantage of being fairly recent and containing discussions of previous remarks and proposed changes .
my own version of it follows now and i try to summarize the useful discussions i had with wojciech de roeck ; let me cut the story in pieces : 1 . first there is the condition of unitarity which is emphasized .
that is not much more than to say we want a description starting from microscopic mechanics .
whatever happens in the formation process or evaporation process of a black hole , the evolution is unitary , mapping the initial wave function to intermediate and final wave functions with a so called unitary @xmath11-matrix .
the unitarity is especially important here because it will lead to the mathematical identity essential to the paradox .
we make the usual splitting of hilbert spaces between the singularity and the outer black hole ; we are just considering now the process of evaporation and radiation being emitted by the black hole .
so we will consider ( for short ) the inner ( singular ) and the outer part of the black hole .
there is the interior of the black hole ( behind the horizon , region b ) and there is the outside or the exterior ( region a ) . the total system ( a and b together ) are quantum mechanically described by a pure state , a wave function @xmath20 .
it is unitarily evolved from another pure state @xmath21 where we put time zero at the beginning of the evaporation process as described by hawking radiation .
if we want to consider the outside region a , we can integrate out the degrees of freedom in b. in quantum mechanics , that means we trace out , and , in contrast to classical states , we do not obtain in general another pure state describing the situation in b , but we obtain a density matrix @xmath22 similarly , there is a density matrix @xmath23 for the statistical distribution of the interior of the black hole .
the fact that these are density matrices ( and not wave functions ) arises because the pairs of photons that are created at the horizon are entangled .
they create in other words an entanglement between regions a and b. since the state @xmath20 was pure , it is a theorem that the entanglement entropy of region a equals that of region b. mathematically , that is an equality between von neumann entropies @xmath24 at all times @xmath25 .
the von neumann entropy of the density matrix in the outer region must be equal to the von neumann entropy of the interior black hole .
3 . consider the entanglement entropy of the outer region a ( left - hand side in ) .
can we estimate that ? here is an important statement
let us call it the statement of _ increased entanglement _
: entanglement just increases with further pair creation .
the reason and calculation which is given of that _ increased entanglement _ is that ( a ) the hawking radiation is thermal , and ( b ) the radiation is additive in pair creation .
an observer external to the black hole will see a thermal state characterized by a density matrix and moreover purity of that state is never to be restored even when all the black hole is evaporated .
the computation of increased entanglement is based on that thermal equilibrium distribution for black body radiation ( at the hawking temperature ) .
we are now at page 17 in @xcite . in the words of the `` fuzzball person '' @xcite : _
thus the entanglement can not go down ever , and thus the information can not emerge in the hawking radiation . _ + the increase of the left - hand side of is taken from the thermal entropy , which is then increasing alright with every creation .
a detailed calculation can be found on pages 9093 in the book @xcite in the paper _ the firewall phenomenon _ of r.b .
mann . to be sure
, there is an upper bound on that entanglement entropy , at first growing linear in the steps of pair creation , and then saturating .
( in @xcite bekenstein showed that there is an upper limit on the amount of entropy ( and thus information ) one can store in a
chunk of space - time " of a certain radius ; a time - dependent version is found in @xcite . )
also , there can easily be imagined corrections to thermality , but one shows that the increase of entanglement is stable ; see again p93 in @xcite , or p6 - 8 in @xcite based on @xcite .
+ secondly and moreover , the calculation for increased entanglement uses that the cumulative pair creation works additively in the entanglement entropy .
every new pair creation adds an elementary unit to the entanglement entropy .
4 . meanwhile , what happens to the region b ?
well , by the hawking radiation the black hole evaporates and its entropy starts to become smaller and smaller .
the radiation in a grows and region b gets smaller . at a certain moment ( beyond the so called page time ) , the region b is getting smaller and smaller and therefore its entanglement entropy which is always smaller than its real ( thermodynamic ) entropy must also decrease to zero .
if there is almost nothing left of the black hole , then its entropy gets very small and hence the right - hand side of must decrease to zero .
5 . now comes the paradox . in the equality ,
the previous lines just showed the right - hand side goes to zero .
but then the entanglement entropy in region a also decreases in time @xmath25 which contradicts the _ increased entanglement _
, that the radiation remains entangled , that the density matrix @xmath26 never restores purity as was inferred before from taking it thermal .
a large number of `` solutions '' and `` answers '' have been proposed ; too many to include all of them .
i start by mentioning some of them very briefly ; some are rather involved and sometimes get very technical .
compared with the simple line of arguing above , in general they appear somewhat off target .
+ the whole description is of course based on calculations that are approximate .
hawking s calculation that showed that black holes emit thermal radiation @xcite uses a semi - classical calculation , say quantum field theory with curved background ( in fact , quantum theory of a scalar field in the background of a large classical black hole ) .
moreover the calculation uses locality for example in assuming the decomposition of the hilbert space structure corresponding to regions a and b. nevertheless i would be very surprised that these assumptions and approximations lead to such disaster .
after all , the size of the horizon can be arbitrarily weakly curved when the mass of the black hole is large enough .
i do not believe that quantum gravity effects can be relevant at such large length scales .
+ what is probably more valid if one wants to criticize these effective theories , is that we had better used _
open effective field theory _ ,
i.e. , effective field theory of systems that are not closed .
of course , then one must explain well the ultimate and cogent reasons for using these open system theories , but that is basically a problem of statistical mechanics we are used to and some elements of it will be discussed in section [ thm ] .
+ the suggestion or solution of `` forgetting degrees of freedom '' has been entertained in e.g. @xcite ; see also @xcite , in which samir mathur explains why some common beliefs do not resolve the apparent puzzle , which sharpens significantly the nature of the information paradox as originally stated by hawking .
it appears that the solution of the quantum information paradox is not to be found in these ideas of `` approximate '' thermalization , as i already have mentioned in the point 3 above for _ increased entanglement_. + a further objection to the paradox ( again to come back in section [ thm ] ) states that the description of the black hole space - time by the schwarzschild metric is truly based on an exchange of limits .
a collapsing shell of matter only becomes schwarzschild in an infinite time limit .
at all finite times it slightly differs .
still i do not believe that it is the good answer to the firewall phenomenon described above .
the solution is much simpler to state .
+ let me no longer postpone giving the answer of th .
smiths : the thermality ( approximate or not ) of hawking radiation does imply the _ increased entanglement _ nor does the cumulative nature of pair creation . just from rereading the scenario in 5 acts of section [ form ]
it is quite clear that the statement of _ increased entanglement _ must be very false that is a matter of logic if one accepts all the rest .
the entanglement entropy just goes to zero , indeed both right - hand side and left - hand side of .
so the arguments that lead to that presumed ( entanglement ) entropy increase is wrong .
why , how so ? + indeed
, even a pure state can very well look like a thermal state for all practical purposes and for local observations ; the von neumann entropy or the shannon entropy are not continuous ; see e.g. @xcite .
it is not because two density matrices very much resemble each other locally , that their von neumann entropies can not be drastically different , and the usual argument for _ increased entanglement _ is just and only based on that wrong premise .
+ the real issue thus concerns the nature of the thermal description .
what is really the meaning and the status of these density matrices and probabilities that seem to enter our description of being thermal ? how seriously should we take them ?
the answer is , not too much . to give a trivial example : the liouville equation for a mechanical system implies that the shannon von neumann entropy of a density matrix ( or phase space distribution ) does not change in time , even though it can on the appropriate space time scale or for some class of observables be considered thermal .
so it then describes equilibrium for many practical purposes and yet its shannon - von neumann entropy ( like in ) does not at all equal the thermodynamic entropy ( which itself however can be written as the shannon - von neumann entropy of a thermal density matrix ) ; see appendix [ sha ] for a toy - example .
that density matrix ( solution of the liouville - von neumann equation ) is for example also quite irrelevant for a more microscopic understanding of the second law .
similarly , a wavefunction for a many - body system can statistically reproduce a thermal state when evaluated for local observables , but that does not imply that the two are equal in all possible senses . certainly , their ( global ) von neumann entropy need not be equal .
one should realize here that a wavefunction or a density matrix for a quantum many - body system contains much more information than for example the one - particle statistics .
so it is not because you can reproduce locally the black body radiation spectrum for a big density matrix that it would entail that its von neumann entropy is even approximately equal to that of the corresponding thermal state .
mathematically , entropy is not a continuous functional ( with respect to such weak metric ) .
+ moreover , even when not inserting ( wrongly ) the `` mathematically thermal '' condition ( i.e. , even when not literally using thermal density matrices ) , one must still avoid a second mistake : that the pair creation as such which is additive , need not lead to increased entanglement .
the reason is simply that the strict correlations with the inner degrees of freedom in the black hole easily get lost ; see appendix [ bag ] for a simple toy - example .
+ the previous answer is just a detection of where was the mistake in the reasoning of section [ form ] . to th .
smiths it would be like saying that the shannon entropy of the liouville evolved probability distribution equals the ( real ) thermodynamic entropy .
that is certainly false , even though there may appear good reasons to say that the distribution is thermal indeed .
hawking radiation is essentially pure but that does not require a breakdown of effective field theory .
nevertheless that answer does not describe the mechanism of purification or how to calculate the true and correct entanglement entropy .
that is a much more detailed and complicated question , which is typically not even tackled for much simpler systems .
but it also seems to take us in the wrong direction , the quantum - mechanical description of black holes and of singularities probably requires much more interesting and important challenges , and it may well be that the answer depends strongly on the version of quantum mechanics that one considers , @xcite .
not so very long ago , speaking of entanglement was considered ( bad ) philosophy @xcite . in those times , the information paradox was formulated quite differently from what is written above , but still with a similar flair as people have in general had less reservations to associate the word information with entropies .
+ hawking s theorem in @xcite showing that black holes emit thermal radiation is then combined with another important fact about classical black holes , namely the no - hair " theorem : _ a stationary four - dimensional solution of the einstein - maxwell equations ( in lorentzian signature ) is uniquely characterized by its mass ( @xmath27 ) , angular momentum ( @xmath28 ) , electric charge ( @xmath29 ) , and magnetic charge ( @xmath30 ) . _
see the reviews @xcite
. + the thermal radiation by black holes together with the no - hair " theorem seem to suggest that one can take a pure state with charges @xmath31 , evolve it in time to form a black hole and then observe a thermal radiation coming out of the black hole .
that is in a nutshell what was the original formulation of the `` information paradox '' since it seems to suggest that one can evolve a pure state into a thermal one thus breaking unitarity .
( the version of the paradox presented above in section [ form ] is more specific and more recent . ) + also here ( and in relation with the firewall phenomenon ) various `` answers '' have been formulated .
there is for example the claim ( contained e.g. in @xcite ) that using ads / cft ( a topic that got recently much attention ) one can understand exactly how information leaks out of an asymptotically ads black hole . those answers and in particular those based on ads / cft dualities may very well be correct , but is there not a much simpler issue here , which is felt intuitively by th .
smiths ? + the reply here must be that the no - hair theorem already supposes a stationary limit - situation .
but in such limiting regimes dissipative effects easily arise and are compatible with pre - limit unitary evolution .
very often , to make things very sharp , we take some thermodynamic limit after which the ( reduced ) description is particularly simple .
it can for example suffice to give energy , density and volume to completely describe a gas , or to give temperature and magnetization to describe a magnet etc .
that is similar to the status of these no - hair theorems in black hole physics : they involve limiting procedures and considerations of asymptotic stationary behavior both in time and in degrees of freedom . in these same limits and for the appropriate variables ,
autonomous dissipative evolutions appear rigorously as mesoscopic and macroscopic behavior with no other input than hamiltonian or unitary microscopic laws .
the first example of that was probably the proof of the boltzmann equation for a dilute gas @xcite .
perhaps that is related to the fuzzball proposal for black holes @xcite .
the basic idea is that the black hole has microstates to account for its thermodynamic entropy and the typical microstate is some fuzzy " space time which has features on the scale of the horizon and at asymptotic infinity looks like minkowski space .
one can then somehow average out over these fuzzy " geometries and obtain the black hole as a coarse - grained effective description of the physics .
+ perhaps we are witnessing something similar to what happened with classical mechanics when confronted with hydrodynamics and thermodynamics . here are two historical examples both concerned with theorems , mathematical rigorous work that has lead to paradoxes that are however solved by understanding their complete irrelevance for the true situation ( which mathematically amounts to exchanges of limits ) . + the first one dates from 1752 and contains what became known as the dalembert paradox @xcite , meaning the rigorous conclusion from classical mechanics that birds can not fly .
( more precisely , dalembert saw that both drag and lift are zero in potential flow which is incompressible , inviscid , irrotational and stationary . )
the resolution is of course found in the emergence of viscosity either from hamiltonian mechanics or from eulerian hydrodynamics , as explained through the validity of the navier stokes equation ( 18221845 ) , or more generally from microscopic ( statistical ) derivations of the second law ( as i had a chance to hint at in section [ 2nd ] ) . + the second one is a paper in 1889 by henri poincar where he shows that no monotonically increasing ( entropy ) function in time could be defined in terms of the canonical variables in a theory of @xmath2-body hamiltonian dynamics @xcite . in that way he added to the so called irreversibility paradox , and
indeed its solution shows that that poincar theorem is quite irrelevant ; see for example @xcite .
to what comes of the horizon problem and the information paradox to the streets of statistical mechanics , there appears the following reply : + for the horizon problem : seeing equilibrium aspects such as homogeneous temperature of the background radiation can not be a problem if there is no reason to suppose it was different before .
the gravitational degrees of freedom appear very far from equilibrium , yet not contradicting the large scale almost equal temperatures without previous causal contact .
+ for the information paradox : unitary evolutions become effectively dissipative in a reduced description , but formal paradoxes can easily arise when interchanging limits of time and thermodynamic limit .
macroscopic steady state descriptions are by their nature approximations , valid in some limiting regimes of spatio - temporal scales . that a statistical distribution is thermal for local observations or for
all practical purposes does not need to imply that its shannon von neumann entropy coincides with that of the thermal distribution .
in particular , the statement that the hawking radiation `` is '' thermal , does not include that the entanglement entropy as seen by the outside observer is the bekenstein hawking entropy or is not ultimately decreasing to zero .
moreover , the cumulative or additive pair creation does also not imply a never - decreasing entanglement entropy of the radiation as the inner degrees of freedom of the black hole , while shrinking , can become much more internally correlated .
( see some toy - examples in the appendices . )
+ there have been times of great mutual interactions between researchers in statistical mechanics and in field theory .
several ideas on dynamically broken symmetries and on collective phenomena have been shared and they shaped the landscapes of condensed matter and elementary particle physics alike .
the renormalization group , universality , effective coupling , etc .
remain key - concepts in all of theoretical physics .
today there appear new opportunities , also for th .
smiths , for joined initiatives and efforts at institutes for theoretical and mathematical physics , including work and discussions on a fluctuation theory for gravitation with special challenges regarding the statistical mechanics of the big bang and black holes .
+ * acknowledgment * + 99 e. schrdinger , _ nature and the greeks_. cambridge university press , 1996 .
f. bouchet and j. barr , statistical mechanics of systems with long range interactions .
journal of physics : conference series * 31 * , 18-26 ( 2006 . ) m. k .- h .
kiessling , the `` jeans swindle '' : a true story mathematically speaking .
advances in applied mathematics * 31 * , 132-149 ( 2003 ) .
the wikipedia initiative is often a source of good information and is very much appreciated .
i quote it here simply because it thus represents the majority of writings on the subject .
it also avoids embarrassing specific authors .
what is `` usual '' is possibly subject of discussion , as it is somewhat subjective .
smiths probably heard the formulation at talks and in popular accounts motivating inflation .
there are of course also a wealth of professional texts saying the same thing . as a recent reference ,
see for example + d. baumann and l. mcallister , _ inflation and string theory_. cambridge university press , 2015 .
see the highlights of the wilkinson microwave anisotropy probe webpage http://wmap.gsfc.nasa.gov/universe/bb_cosmo_infl.html last visited 1 april 2015 , with the text : + _ the horizon problem : distant regions of space in opposite directions of the sky are so far apart that , assuming standard big bang expansion , they could never have been in causal contact with each other .
this is because the light travel time between them exceeds the age of the universe . yet
the uniformity of the cosmic microwave background temperature tells us that these regions must have been in contact with each other in the past .
+ since inflation supposes a burst of exponential expansion in the early universe , it follows that distant regions were actually much closer together prior to inflation than they would have been with only standard big bang expansion .
thus , such regions could have been in causal contact prior to inflation and could have attained a uniform temperature . _
there are other claims and formulations under the horizon problem , see again @xcite .
a popular emphasis is on the high correlations between causally not connected regions or between temperature and other physical fields .
i have not been able to find a precise formulation of that `` correlation - paradox '' which is not either the same as the `` more or less '' equal temperature problem , or which is a problem at all ( given what one means by correlations which of course need not be causal in general ; apple trees and pear trees do not cause their blossoming but the time of their flowering is about equal . )
r. penrose , _ the road to reality : a complete guide to the laws of the universe_. vintage ; reprint edition ( january 9 , 2007 ) .
carroll , in what sense is the early universe fine - tuned ?
arxiv:1406.3057v1 [ astro-ph.co ] . + that paper also contains the complete solution of the so called flatness problem , another historical motivation for inflation . an earlier paper exactly hitting that target is + g. evrard and p. coles , getting the measure of the flatness problem . class . quantum grav . *
12 * , l93-l97 ( 1995 ) .
s. weinberg , gravitation and cosmology .
new york , wiley , 1972 .
+ `` unfortunately , we still do not have even a tentative quantitative theory of the formation of galaxies ... '' r. tolman , _ relativity , thermodynamics and cosmology_. ( dover books on physics ) january 20 , 2011 ; clarendon press oxford , 1934 . for example , it is not unusual to hear or read that the paradox is about information , with the key - question whether hawking radiation carries information or not .
another emphasis regards the paradox as a contradiction between predictions of semi - classical gravity and expectations for a ( good ) quantum theory of gravity . or still others would speak about the unwanted pure to mixed transition ; the hawking radiation would become mixed / thermal because of paired particles disappearing in the singularity , thus erasing information , etc .
all of these concerns are recognized and more sharply stated in the recent debates on fire walls .
we come to a more precise formulation in section [ form ] .
l. susskind and j. lindesay , _ an introduction to black holes , information and the string theory revolution .
the holographic universe_. world scientific 2005 . , ed .
xavier calmet , springer fundamental theories of physics * 178 * , 2015 . in particular , see the paper on the firewall phenomenon therein , by r.b .
a. almheiri , d. marolf , j. polchinski and j. sully , black holes : complementarity or firewalls ?
journal of high energy physics * 2 * , 62 ( 2013 ) . the webpage of samir d. mathur has many entries on the black hole information paradox including a pedagogical blog .
( last visited 9 april 2015 ) .
on page 5 of `` confusions and questions about the information paradox '' by samir d. mathur ; see @xcite .
a. almheiri , d. marolf , j. polchinski , d. stanford and j.sully , an apologia for firewalls . arxiv:1304.6483v2 [ hep - th ] .
mathur , what the information paradox is _
not_. arxiv:1108.0302 . s.d .
mathur , the information paradox : a pedagogical introduction .
* 26 * , 224001 ( 2009 ) .
s. w. hawking , particle creation by black holes .
phys . * 43 * , 199 ( 1975 ) .
[ erratum - ibid . *
46 * , 206 ( 1976 ) ] .
j. d. bekenstein , a universal upper bound on the entropy to energy ratio for bounded systems .
d * 23 * , 287 ( 1981 ) .
r. bousso , a covariant entropy conjecture .
jhep * 9907 * , 004 ( 1999 ) .
dobrushin , a mathematical approach to foundations of statistical mechanics .
boltzmann s legacy 150 years after his birth ( rome , 1994 ) , atti covegni lincei , 131 , accad .
lincei , rome , 1997 , 227243 .
f. t. falciano , n. pinto - neto and w. struyve , wheeler - dewitt quantization and singularities .
d * 91 * , 043524 ( 2015 ) .
i remember that was still often the case in the 1990s ; entanglement was at the very best considered to be some philosophical term for what was obvious and not so interesting after all .
it remains mind - boggling that entanglement was introduced already by schrdinger in the 1930 s , that john bell used it essentially in his analysis of the bohm - epr experiment , and that still today simple understanding of the content of e.g. the bell inequality ( 1964 ) has remained largely lacking or gets confused with other statements as the kochen - specker theorem . for a good review , see + t. maudlin , what bell did .
j. phys . a : math
. theor . * 47 * , 424010 ( 2014 ) , part of the special issue of journal of physics a : mathematical and theoretical devoted to `` 50 years of bell s theorem . ''
the rigorous proof of the boltzmann equation in the so called grad - limit , at least for short times but without explicit assumptions of molecular chaos ( the sto@xmath33zahlansatz ) was given in + o.e .
lanford iii , the evolution of large classical system . in : _ dynamical systems ,
theory and applications _
, j. moser , ed . ,
lecture notes in physics 38:1 111 , springer - verlag , heidelberg , 1975 .
+ see also www.scholarpedia.org/article/boltzmann-grad_limit ( last visited 9 april 2015 ) .
s. d. mathur , the fuzzball proposal for black holes : an elementary review .
fortsch .
phys . * 53 * , 793 ( 2005 ) .
b. d. chowdhury and s. d. mathur , radiation from the non - extremal fuzzball .
grav . * 25 * , 135005 ( 2008 ) .
t. chrusciel , no hair theorems : folklore , conjectures , results .
contemp .
math . * 170 * , 23 ( 1994 ) .
robinson , four decades of black hole uniqueness theorems . in _ the kerr spacetime : rotating black holes in general relativity _ , edited by d. wiltshire et al . , cam- bridge university press , 2009 .
jean le rond dalembert , essai dune nouvelle thorie de la rsistance des fluides .
( 1752 ) .
h. poincar , sur les tentatives dexplication mcanique des principes de la thermodynamique .
comptes rendus hebdomadaires de lacadmie des sciences de paris * 108 * , 550553 ( 1889 ) .
s. goldstein , boltzmann s approach to statistical mechanics . in _ chance in physics : foundations and perspectives _ , edited by jean bricmont , detlef durr , maria c. galavotti , giancarlo ghirardi , francesco petruccione , and nino zanghi , lecture notes in physics 574 , ( springer - verlag ) 2002 .
w. de roeck , t. jacobs , c. maes and k. neton , an extension of the kac ring model .
gen . * 36 * , 113 ( 2003 ) . c. maes , k. neton and b. shergelashvili , a selection of nonequilibrium issues . in : _
methods of contemporary mathematical statistical physics _
roman koteck , lecture notes in mathematics 1970 , pp .
247 - 306 , springer , 2009 .
i repeat here some ingredients of the quantum kac ring model @xcite to show how a quantum unitary evolution gives rise to `` effective '' dissipative behavior for some thermodynamic observables .
in fact the model has similar semi - classical aspects as found for quantum field discussions in a curved back ground .
yet , the kac model is so transparent that all calculations become very simple .
+ consider a ring @xmath34 of a large number @xmath2 of sites each carrying a quantum spin @xmath35 and a classical background variable @xmath36 .
we characterize that background by the average @xmath37 . for the quantum dynamics
we choose a unitary matrix @xmath1 on @xmath38 and define @xmath39 @xmath40 which is extended by linearity to be unitary on the hilbert space @xmath41 ( given the background @xmath42 which is unchanged in the dynamics ) .
+ for observables we take one - site matrices , the pauli matrices @xmath43 in the three directions @xmath44 and magnetization operator @xmath45 where the @xmath46 are copies of @xmath43 at site @xmath47 . as initial condition we can essentially take any situation which is concentrating on a particular magnetization vector @xmath48 , by which we mean an initial state @xmath49 for which the expectations @xmath50 \rightarrow m_\alpha$ ] for large @xmath2 ( convergence in mean ) . in other words , magnetization is initially well - defined .
we now evolve unitarily @xmath51= \omega^n\left[(u_n)^{-t}\,\cdot\,(u_n)^t\right ] \quad \text { from } \omega^n_0=\omega^n \quad \text { with } t=0,1,2,\ldots\ ] ] and we are interested in the large @xmath2 statistics .
remember that @xmath49 has magnetization @xmath48 .
one can then prove @xcite that , as @xmath52 @xmath53 \rightarrow f((\phi_tm)_\alpha ) , \quad \alpha=1,2,3\ ] ] for all continuous functions @xmath54 , where @xmath55 is a ( truly-)dissipative autonomous dynamics in the magnetization - space .
in other words , the unitarily evolved quantum state @xmath56 can be seen as the result of a dissipative evolution when restricted to the magnetization observables . for large @xmath2 and any fixed time @xmath25 the induced statistics on
the magnetization is indistinguishable from that obtained from a relaxational dynamics towards equilibrium .
+ we can still say it differently .
any magnetization vector @xmath48 gives rise to a @xmath57 density matrix @xmath58 .
we define the dissipative evolution , with previously introduced background parameter @xmath59 , @xmath60 which after @xmath25 steps takes @xmath61
. then @xmath62 $ ] .
concerning magnetization observations the theorem thus states that for all times @xmath63 , disordered with a magnetization evolving dissipatively .
+ obviously , the von neumann entropy of @xmath64 is constant in time @xmath25 , but that of @xmath65 is not .
understandably , the tr@xmath66 $ ] expression for @xmath64 and for @xmath67 differ enormously .
the latter is truly thermal for large time @xmath25 with a temperature that is directly derived from the initial condition , and its boltzmann entropy , gibbs entropy , von neumann entropy and thermodynamic entropy all coincide .
nothing like that is true for @xmath56 and yet , it is `` thermal '' as well for all purposes of magnetization .
the point of the following little story is to show that even when we enforce strict thermalization ( in a mathematical sense ) , still the correlations need not keep increasing . here
we thus leave quantum unitary evolutions , possibly having in mind an effective description on a subclass of observables as in the previous appendix [ sha ] , and concentrate on the relation between additive pair creation and building up correlations .
+ suppose we have a big container with a total of ( a very large even number ) @xmath2 black and white balls .
they are moving around quite chaotically in a container with a small hole leaving room for one ball at a time to escape the container .
let us count time discretely with one step containing two escapes .
( the escapes mimic the pair creation . ) at moment @xmath68 a total of @xmath69 balls went through that hole .
it is organized in such a way that at their escape ( say at the hole ) another chaotic control repaints the balls either black or white with equal probabilities but so that necessarily , within each time step , the two escaping balls get opposite colors . ( the pair escaping at time @xmath70 has perfect correlation . ) finally , for each escaping pair one ball is returned inside the container and the other ball is collected in a bag .
all that is classical mechanics and there will be no talk of entanglement ; rather we speak here about the correlation between the average color in the bag and the average color in the container . yet
, very similar models can be set up also for quantum unitary evolutions where the correlation is truly measured in terms of entanglement ( but not here in the present paper ) .
+ for the purpose of calculation we thus have @xmath71 at time @xmath70 for the balls in the container .
we take for example black to be + 1 and and we call -1 white .
we start at time zero with an equal number of black and white balls . the total color in the container at time @xmath70 is @xmath72 ( @xmath73 . )
the @xmath74label has no permanent meaning ; it just orders the balls at each moment . at all times
the bag is as disordered as as series of coin tossing ; balls that fell in the bag are black or white , independently and with equal probability .
( they are the analogue of thermal radiation . ) yet the bag s color is correlated with the container s color , surely increasingly so in an initial period .
+ let @xmath75 denote the colors of the balls in the bag , with total color @xmath76 ( and here the @xmath77label can denote the order in which the balls arrived in the bag , but that is of no further importance ) . at time
zero @xmath78 say ( no balls in the bag ) . at time
@xmath79 one ball @xmath80 arrived in the bag , black or white with equal probability , and in the container @xmath81 has three possible values , @xmath82 with probability 1/2 , 1/4 and 1/4 respectively . at any event , @xmath83\,\xi$ ] where @xmath84 is the color of the ball that was returned inside the container , with @xmath85 with equal probability . at that moment the balls in the container
are reshuffled and again two of them are picked out , colored randomly but differently after which one returns to the container and the other is added to the bag , _ etcetera_. the correlation at time @xmath70 when averaged over many runs equals @xmath86 .
there is an initial period ( till about time @xmath87 ) of increased ( anti-)correlation after which , time - symmetrically , ( anti-)correlations decrease .
+ i repeat that the present example is not taking care of unitarity ; it just makes the point that ( 1 ) one can easily produce maximal disorder ( in the bag ) , and additively in ( 2 ) each time , for each pair , impose strict anti - correlation , while still ( 3 ) the correlation is non - monotone in time with the source .
i thank urna basu for discussions on that toy - model . | e tatistical echanician n e treet ( our th .
smiths ) must be surprised upon hearing popular versions of some of today s most discussed paradoxes in astronomy and cosmology .
in fact , rather standard reminders of the meaning of thermal probabilities in statistical mechanics appear to answer the horizon problem ( one of the major motivations for inflation theory ) and the information paradox ( related to black hole physics ) , at least as they are usually presented .
still the paradoxes point to interesting gaps in our statistical understanding of ( quantum ) gravitational effects . |
effective potentials are a key to unlocking the equilibrium behavior of many soft - matter systems@xcite .
the basic philosophy behind this coarse - graining approach is that the initial effort in deriving these potentials is recouped when they are input into the well oiled machinery of liquid state theory@xcite , or when they are used in computer simulations@xcite .
an archetypal example is the depletion potential , induced between colloidal particles by non - adsorbing polymers .
asakura and oosawa@xcite first showed that a bath of such polymers , characterized by their radius of gyration @xmath0 , induces an attractive depletion interaction of range @xmath1 between two plates .
their calculation was exact for non - interacting polymers .
later , the same authors , and independently vrij@xcite , derived a depletion potential between two colloidal hard spheres ( hs ) by approximating the ( ideal ) polymers as penetrable spheres .
this is often termed the asakura - oosawa ( ao ) model .
a good example of the effective potential coarse - graining approach is the calculation of the phase behavior of polymer - colloid mixtures by gast , hall , and russel@xcite , and also by lekkerkerker _
et al._@xcite and meijer and frenkel@xcite
. they found , using an ao depletion potential approach , that the fluid - fluid phase line of colloids of radius @xmath2 becomes metastable w.r.t . the fluid - solid phase line if the size ratio @xmath3 is less than about @xmath4 , in qualitative agreement with experiments@xcite .
their work demonstrates how an accurate knowledge of the depletion potential can lead to a good understanding of the equilibrium behavior of colloid polymer mixtures .
the latter are important not only because of their relevance to many industrial and biological processes , but also because they form an important model system for equilibrium and non - equilibrium behavior in soft matter science . whereas the depletion interaction for ideal polymers is now quantitatively understood ,
the depletion interaction induced by polymers with excluded volume interactions is only qualitatively understood .
experiments on the phase behavior@xcite and structure@xcite of polymer colloid mixtures also show deviations from the simple ao model@xcite , as do direct measurements of the depletion potentials@xcite .
theoretical attempts to directly calculate the depletion potentials for interacting polymers include scaling theory@xcite , self - consistent field theory@xcite , perturbation theory@xcite , direct simulations@xcite , rg theory@xcite , as well an interesting new `` overlap approximation '' method@xcite .
all these approaches show significant deviations from ideal polymer behavior , but many questions still remain .
this is in contrast to binary hs colloid mixtures , where the deviations from the ao potential can now be quantitatively calculated with density functional theory ( dft)@xcite , and the effects of non - ideality on the phase behavior are fairly well understood@xcite .
the goal of our paper is to derive a theory of similar accuracy for the depletion potential induced by interacting polymers in a good solvent before we proceed , an important caveat is that the depletion potential becomes less relevant for phase behavior at larger size ratios @xmath5 , since many - body interactions become increasingly important@xcite .
when @xmath6 , i.e. when the colloids are much smaller than the polymers , other approaches , which treat the polymers on a monomer level , are more relevant .
examples include integral equation techniques@xcite , scaling theory@xcite , or renormalization group theory ( rg)@xcite . in this paper
we concentrate on the regime where @xmath0 is of the order of @xmath2 or smaller .
we have recently performed a systematic study of the insertion free - energy of a single colloidal particle in a bath of interacting polymers@xcite ( henceforth referred to as paper i ) , and found important deviations from ideal polymer behavior . in particular
, we found that the range of the depletion layer decreases with increasing density @xmath7 , and that this effect is most pronounced in the dilute regime , where @xmath8 ; here @xmath9 is the overlap density . in the semi - dilute regime , where @xmath10 , we found that the effects of the interactions could be well described by scaling theory@xcite . since these one - body effects were not well captured by ( non - additive ) hs like models , we do not expect straightforward extensions of the dft techniques that work so well for binary hard hs mixtures , to also perform well for two - body depletion potentials in polymer - colloid mixtures . to calculate the depletion potentials
, we use a similar approach to that used in paper i a combination of scaling theories and computer simulations .
since we found in paper i that the range of the depletion layer decreases with increasing density , we expect a similar trend for the range of the related depletion potential . throughout this work we focus on the dilute and semi - dilute regimes@xcite of the polymers , where the monomer density @xmath11 is low enough for detailed monomer - monomer correlations to be unimportant ; the melt regime , where @xmath11 becomes appreciable ,
will not be treated here . since our models
are all athermal , we set @xmath12 . the paper is organized as follows : the case of the depletion interaction between two plates is discussed in detail in section ii .
we show that it is closely related to the problem of determining the surface tension @xmath13 , which was solved in paper i. just as was found for @xmath13 , the depletion potential simplifies in the semi - dilute regime . in section iii
, we discuss the depletion interaction between two spheres .
we compare the results of direct monte carlo ( mc ) simulations of the interaction between two hs colloids induced by a bath of self avoiding walk ( saw ) polymers to a potential derived within the derjaguin approximation@xcite .
the derjaguin approximation works much better than one would naively expect because of a cancellation of errors related to the deformation of polymers around a sphere . using an extension of the derjaguin approximation
, we derive a new semi - empirical depletion potential which appears to be nearly quantitative for @xmath14 , the regime where depletion potentials are most relevant for the phase behavior and structure of polymer - colloid mixtures .
we derive the scaling behavior with density , and find important deviations from the ao model and other ideal - polymer theories .
we also calculate the depletion potential between spheres within our new `` polymers as soft colloids '' coarse - graining scheme@xcite , where each polymer is represented by a single particle , interacting via a density - dependent effective potential . here
we again find good agreement with the direct mc results for the dilute regime , but for @xmath15 , significant deviations are found . in section iv
, we discuss the effect of polymer density on the virial coefficients between two colloids , and relate this to phase behavior of polymer - colloid mixtures .
immersing a single hard wall ( or plate ) into a bath of non - adsorbing polymers in a good solvent reduces the number of configurations available for the polymers .
this in turn results in an entropically induced depletion layer @xmath16 near the wall , discussed in more detail , for example , in paper i. associated with the creation of this depletion layer is an interfacial free energy cost per unit area @xmath17 , i.e. a wall - fluid surface tension @xmath13 , which typically depends on the bulk density @xmath7 , or equivalently the bulk chemical potential @xmath18@xcite .
bringing two such walls , of area @xmath17 , together from an infinite distance apart to a distance @xmath19 where the two depletion layers begin to overlap , changes the total free energy of the polymer solution .
this change in the free energy or grand potential @xmath20 per unit area is called the depletion potential @xmath21@xcite . when @xmath22 , i.e. when the two walls are brought into contact , the depletion potential reduces to the simple form @xmath23 reflecting the fact that the two depletion layers are completely destroyed . in paper
i , we used an extension of the gibbs adsorption equation to express the surface tension near a single wall as @xmath24 the derivation of this equation can be found , for example , in@xcite . here @xmath25 is the osmotic pressure of the polymer solution , and @xmath26 is the reduced adsorption near a single wall , defined as : @xmath27 where @xmath28 is the excess grand potential per unit area . in paper
i , we used computer simulations of a saw polymer solution in a good solvent near a wall to calculate @xmath16 and @xmath26 for several values of @xmath29 .
as discussed in paper i , the saw on a cubic lattice is a very good model for polymers in a good solvent . in the scaling limit , where the length @xmath30 tends towards @xmath31 ,
its properties are universal , and agree with experiments on polymers in the same good solvent regime@xcite .
for example , the radius of gyration scales as @xmath32 , where @xmath33 is the flory exponent , taken to be @xmath34 in this paper . by using an accurate fitting form which takes into account the correct scaling behavior , we expressed @xmath26 for all densities in the dilute and semi - dilute regimes . when this was combined with an rg expression for the pressure
, we were able to calculate the surface tension @xmath13 as a function of density through eq .
( [ eq2.2 ] ) .
our results agreed very well with some recent rg calculations@xcite .
here we use @xmath13 together with eq .
( [ eq2.1 ] ) to directly calculate the depletion potential at contact , as shown in fig .
[ fig : w02-w0semi ] .
these results are also compared to direct computer simulations of @xmath35 saw polymers@xcite . even though the @xmath35 simulations have not quite reached the scaling limit , the well - depths at contact
are remarkably well reproduced , a result also found by tuinier and lekkerkerker@xcite with a similar theory .
the simplest approximation for the depletion potential at finite separation @xmath19 would be a linear form with a slope equal to the osmotic pressure .
this follows because @xmath36 is the volume term proportional to the work per unit area produced by the osmotic pressure : @xmath37 where the range is given by @xmath38 we note that this approximation is similar to that adopted by joanny , leibler , and de gennes who , in their pioneering paper@xcite , approximated the force between two plates as constant for @xmath39 and zero for @xmath40 , where @xmath41 is the correlation length , the relevant length - scale in the semi - dilute regime@xcite .
this also results in a linear depletion potential . in fig .
[ fig : vwall - simple ] we compare our simple linear potential to the direct saw simulations .
the overall agreement is striking .
the only ( small ) deviation occurs at larger distances @xmath19 where the true potential rounds off and develops a very small maximum before going to zero@xcite .
the range @xmath42 of this simple potential decreases with @xmath29 .
in fact , as shown in fig . [
fig : range ] , the largest relative rate of decrease occurs in the dilute regime , so that at @xmath43 , the range is @xmath44 , about @xmath45 of the low density limit of @xmath46 . in general , the range of a potential is not always unambiguously defined@xcite .
of course for our linear one it is , and this simple definition would seem a very reasonable definition for the range of the full depletion potentials depicted in fig .
[ fig : vwall - simple ] .
further simplifications occur in the semi - dilute regime .
for example , when the scaling forms@xcite for the osmotic pressure , @xmath47 and the reduced adsorption @xmath48 are used in eq .
( [ eq2.2 ] ) , then , as shown in paper i , the integral simplifies and the potential at contact takes on the very simple form : @xmath49 the linear depletion potential ( [ eq2.6 ] ) then reduces to : @xmath50 and the range has simplified to @xmath51 as can be seen in figs [ fig : w02-w0semi ] and [ fig : range ] , these simple expressions for the well - depth and the range work remarkably well in the semi - dilute regime , especially for @xmath15 . deviations
do occur for low densities since according to ( [ eq2.1 ] ) and ( [ eq2.2 ] ) , @xmath52 for @xmath53 , the same form as for ideal polymers .
( [ eq2.7 ] ) therefore overestimates the well depth by a factor @xmath54 , as can be seen in the inset of fig .
[ fig : w02-w0semi ] . in the same limit the range of the depletion potential reduces to the functional form @xmath55 ,
so that eq .
( [ eq2.9 ] ) also overestimates the range at low densities by a factor @xmath54 , as can be seen in fig .
[ fig : range ] . in the semi - dilute regime
one can identify @xmath56@xcite .
therefore the ansatz @xmath57 , originally postulated by joanny et al.@xcite , is very close to @xmath58 , the expression we derived for the range of the depletion potential in the semi - dilute regime .
their potential is therefore quite accurate in the semi - dilute regime , but overestimates the range and the well - depth by a factor slightly larger than @xmath54 in the dilute limit . for completeness
we compare the results obtained in the previous section to those for ideal polymers , first obtained by asakura and oosawa in 1954@xcite .
their ( exact ) depletion potential can be quite accurately approximated by a simple linear form@xcite : @xmath59 however , for interacting polymers , this is only true in the limit @xmath53 ; the validity of this expression rapidly deteriorates with increasing density . as shown in fig .
[ fig : w02-w0semi ] , eq .
( [ eq2.10 ] ) underestimates the well depth for all but the lowest densities , while , as shown in fig [ fig : range ] , it overestimates the range at all densities .
one might think that replacing the ideal pressure @xmath60 in eq .
( [ eq2.10 ] ) by the pressure of an interacting polymer system would bring an improvement for the well depth , even if this does not improve the approximation for the range . instead , as shown in fig .
[ fig : w02-w0semi ] , this naive approach leads to a severe overestimate of the well - depth .
in fact , for the semi - dilute regime , the contact value of this naive `` improvement '' would scale as @xmath61 instead of the correct @xmath62 scaling .
even for @xmath63 the differences are significant : @xmath64 , while @xmath65 and @xmath66 . for @xmath67 , @xmath68 , @xmath69 and @xmath70 !
in other words , the simple heuristic ansatz is almost never a real improvement ; for interacting polymers , the asakura - oosawa potential between two plates is only accurate at the very lowest of densities .
we now turn to the polymer induced depletion interaction between two hard spheres . similarly to the case of two walls , when the surfaces of two such spheres are brought to within a distance where their depletion layers begin to overlap , the total free energy of the system changes ; this change again defines the depletion potential .
since we found in paper i that the surface tensions associated with the polymer depletion layers around a single sphere show behavior which differs from that of ideal polymers or that of the ao model , we expect to find qualitative differences in the depletion potentials as well . the trends are expected to be similar to those found for the depletion potential between two walls , namely that the range should decrease and the well depth should increase with increasing polymer concentration . to directly calculate the polymer induced depletion potential between two spheres , we performed grand - canonical simulations of @xmath71 saw polymers on a lattice of size @xmath72 , and computed the osmotic pressure or force exerted on two hard spheres placed in the same simulation box .
the configuration space of polymers was sampled in the grand canonical ensemble with a combination of configurational bias monte carlo and pivot moves@xcite .
the depletion force on the spheres was obtained from the ratio of the acceptance of virtual inward and outward moves of the spheres . in fig .
[ fig : forces ] these forces are shown for three different size ratios @xmath73 , namely @xmath74 , @xmath75 , and @xmath76 .
for the first two size ratios , we computed the forces at three densities , @xmath77 , @xmath78 , and @xmath79@xcite , for @xmath76 , this was only done for the lower two densities . as expected , for a fixed size ratio , the range decreases , and the force at contact increases with increasing polymer density .
for a given density , the range appears to contract slightly with increasing @xmath5 , as might be expected , since the polymers can deform more readily around the smaller colloids ( see the appendix for further discussion of this point ) . in each simulation
we keep the size of the polymers fixed , so that the larger size ratios essentially correspond to smaller spheres .
the computational costs scales with the size of the simulation box , which roughly sets the number of polymers needed to achieve a density @xmath29 in the accessible volume left by the colloids .
the force can then be integrated to obtain the effective depletion potential : @xmath80 results are shown in fig .
[ fig : vrad16-ejm - newpot ] for the same parameters as in fig .
[ fig : forces ] .
because the curves are integrated , they appear smoother than the forces .
there may still be some residual error in the potentials , which may explain why the range for the highest density seems slightly larger for the potential than for the force .
also , the errors are likely too large to decide whether or not there is a repulsive bump in the potential .
if it does exist , it is probably very small@xcite .
this is in contrast to hard - core systems , where such bumps can be pronounced@xcite .
the weakness or absence of a repulsive barrier in the case of interacting polymers can probably be traced to the very low monomer concentration @xmath11 . for shorter polymers ,
whose behavior deviates significantly from the @xmath81 scaling limit , and where a substantial monomer density @xmath11 can more easily be achieved , such repulsive bumps might occur more readily@xcite .
as expected , the range decreases while the force at contact and the potential well - depth increase with increasing polymer density . for a fixed density and polymer size @xmath0 , the force at contact and the well - depth decrease as the spheres become smaller ( or @xmath5 becomes larger ) .
the same effect occurs for ideal polymers@xcite , where it has a simple geometric origin : the volume of a depletion layer of a given width @xmath82 decreases with decreasing size of the hard spheres . in the case of two walls , the well - depth at contact
has a clear interpretation in terms of the complete destruction of two depletion layers , and can therefore be expressed as a simple function of the wall - polymer surface tension @xmath13 , as shown in eq .
( [ eq2.1 ] ) . for two spheres ,
the depletion layers do not completely overlap , and the depletion potential at contact is related to the surface tension of polymers surrounding two spheres in a dumbbell type configuration .
a simple expression for the well depth at contact in terms of the surface tension @xmath83 is therefore less obvious .
one way to make contact with the two wall case is to use the derjaguin approximation@xcite , which relates the force between two spheres of radius @xmath2 to the potential between two walls , @xmath84 , in the following way : @xmath85 where @xmath19 is the distance between the surfaces of the two spheres . in principle
this approximation should only be accurate for very small size ratios @xmath5 , i.e. for very large colloids .
as shown in appendix a , we expect there to be a cancellation of errors , related to the deformation of the polymers around spherical particles , which makes the derjaguin approximation work much better than one would naively expect .
this is confirmed in fig .
[ fig : forces ] , where the derjaguin expression for the force , taken from eq .
( [ eq2.6 ] ) and eq .
( [ eq3.2 ] ) , is shown to be a surprisingly good approximation .
it is most accurate for the smallest @xmath5 , as expected , but even for @xmath76 , where normally one would not expect the derjaguin approximation to be useful at all , it is still reasonable . by combining eq .
( [ eq2.6 ] ) with eqs .
( [ eq3.1 ] ) and ( [ eq3.2 ] ) we obtain the following derjaguin expression for the depletion potential between two spheres @xmath86 for @xmath87 ; @xmath88 for @xmath89 . by the nature of the derjaguin approximation
, @xmath42 is the same range as found for two walls in eq .
( [ eq2.6a ] ) .
this simple potential is compared in figs .
[ fig : vrad16-ejm - derj ] to the direct @xmath71 saw simulation results for @xmath75 .
the correspondence is surprisingly good given the large size ratio @xmath5 , although it is not quantitative as in the wall - wall case .
similar results are found for the two other size ratios .
overall , the derjaguin approximation overestimates the well - depth , a consequence of the slightly longer ranged forces found in fig .
[ fig : forces ] .
again , the cancellation of errors found for the simpler ao model in appendix a helps explain why eq .
( [ eq3.10 ] ) works reasonably well in a regime where the derjaguin approximation would normally break down .
we now turn to the scaling behavior of the depletion potential in the semi - dilute regime . by using the simple expression for the range @xmath90 given in eq .
( [ eq2.9 ] ) , the derjaguin approximation for the depletion potential ( [ eq3.10 ] ) reduces to @xmath91 ^ 2\ ] ] for @xmath92 ; @xmath93 for @xmath94 . for two walls , the simplified expression ( [ eq2.9 ] ) for the range
overestimates the true range by a factor @xmath54 in the low - density limit . here
the overestimate is slightly larger , since , as shown in the appendix , the deformation of the polymers around a single colloid reduces the depletion layer width for decreasing sphere size @xmath2 . similarly the well - depth at contact is also overestimated in the low density limit , but now by a factor @xmath95 . in the semi - dilute regime the two expressions come closer for increasing density . for example , at @xmath96 the overestimate at contact is only @xmath97 , while for @xmath98 the two expressions are within @xmath99 of each other .
if in addition we assume that @xmath100 then the expression in eq .
( [ eq3.11 ] ) is again very similar to the form proposed by joanny _
et al._@xcite for two spheres ( @xmath101 being replaced by @xmath102 ) .
our arguments therefore provide an a - posteori justification of the validity of their potential for the semi - dilute regime . in the dilute regime
it will overestimate the attraction in the same way as eq .
( [ eq3.11 ] ) does . given that we now have a reasonably accurate depletion potential between two spheres , namely eq . ( [ eq3.10 ] ) , it seems fruitful to examine how this expression varies with density .
we do this in fig .
[ fig : vderj ] for a number of densities in the dilute and the semi - dilute regimes .
since within the derjaguin approximation the depletion potential ( [ eq3.10 ] ) always scales with @xmath2 , the curves in the plot can be used for all size ratios . of course one
should keep in mind that the derjaguin approximation becomes progressively less accurate for decreasing @xmath103 . with this caveat in mind , the well - depth at contact for polymers in a good solvent scales as : @xmath104 in the semi - dilute regime
this is exactly half the exponent with which the well - depth at contact scales for two walls in the semi - dilute regime ( see eq .
( [ eq2.7 ] ) ) , which can be understood as follows : within the derjaguin approximation the force scales as @xmath105 ; the potential is then obtained through eq .
( [ eq3.1 ] ) by integrating this force over a range @xmath106 .
these two effects result in the scaling seen in eq .
( [ eq3.12 ] ) . on the other hand , by nature of the derjaguin approximation , the range is the same as that shown in fig .
[ fig : range ] for the wall - wall case , i.e. it decreases as @xmath107 .
an expression similar to eq .
( [ eq3.11 ] ) has been derived by fleer , scheutjens and vincent ( fsv)@xcite .
they replace @xmath108 in the full ao potential ( [ eqa.1 ] ) by @xmath109 , and the density @xmath7 by @xmath25 . in its original applications@xcite
, the fsv theory was used with a self consistent field theory to calculate @xmath26 , and a simple flory huggins prescription for @xmath25 .
this leads to @xmath110 scaling behavior which is more appropriate for polymers near the theta point@xcite .
however , one could easily generalize the fsv approach and include the @xmath26 and @xmath25 appropriate for polymers in a good solvent .
for the semi - dilute regime the fsv approach would then lead to an underestimate of the range by a factor @xmath111 , while for small @xmath5 , where the derjaguin approximation is quite accurate , it would underestimate the attraction at contact by a factor of about @xmath112 . on the other hand , in the dilute regime , this approach should perform better , although for increasing @xmath5 it is expected to overestimate the attraction at contact in a very similar way to the effects seen for the ao model in fig . [
fig : vao - derj ] ( a ) .
this is because the fsv model ignores the deformation of polymers near a spherical surface
. it would be better to use the @xmath113 appropriate for a spherical particle ; this results in a potential similar in spirit to an interesting recent proposal by tuinier and lekkerkerker@xcite . as can be seen in fig .
[ fig : forces ] , the derjaguin approximation seems to be particularly accurate for the force at contact .
in fact , when we plot @xmath114 v.s .
@xmath29 in fig .
[ fig : f0 ] , the results for the three size ratios are very close to each other at each of the densities studied , and their value is well described by @xmath115 , as expected for the derjaguin approximation ( see eq . [ eq3.2 ] ) .
furthermore , we analyzed the computer simulation data of dickman and yethiraj@xcite , which used the fluctuation bond model ( fbm ) for size ratios ranging from @xmath116 to @xmath117 , and find similar agreement in fig .
[ fig : f0 ] .
keep in mind that the fbm results are for shorter chains ( @xmath118 ) , and that the error bars appear to be larger than those in our simulations . since for @xmath117 the derjaguin approximation ( [ eq3.2 ] ) should not be valid at all , this does suggest that the expression @xmath119 may hold because of an ( approximate ) sum - rule , and that the agreement is not fortuitous .
comparison of the force at contact with the derjaguin expression from eq .
( [ eq3.2 ] ) .
the circles are taken from the saw simulations depicted in fig .
[ fig : forces ] . on the scale of this graph , the three size ratios lead to virtually identical results at each density
the triangles denote @xmath114 from the fluctuation bond model simulations of yethiraj and dickman@xcite .
the diamonds are from previous @xmath35 saw simulations of @xmath120 , the interaction between two plates@xcite .
the curve denotes @xmath115 , the expected result for the derjaguin approximation .
the surprising accuracy of expression ( [ eq3.3 ] ) for the force at contact can be exploited to derive a more accurate effective potential than that found by the direct application of the derjaguin approximation .
we assume the same linear force between two spheres as found in the previous two sections , but now with a modified range @xmath121 where the effective shrinking of the range due to the deformation of the polymers around the hs has been taken into account via eq .
( [ eqa.4 ] ) .
although this deformation effect is calculated for ideal polymers , we expect it to be a good first approximation for the _ relative _ deformation of interacting polymers around a sphere . the resulting simple linear expression for the force
can now be integrated up via eq .
( [ eq3.1 ] ) to give the depletion potential : @xmath122 for @xmath123 .
@xmath124 for @xmath125 .
our new semi - empirical potential works remarkably well , as can be seen in fig .
[ fig : vrad16-ejm - newpot ] for the three size ratios . even for @xmath76 , corresponding to the smallest colloids , this potential is significantly better than the derjaguin expression ( [ eq3.10 ] ) , suggesting that eq .
( [ eq3.21 ] ) can be considered a nearly quantitative representation of the depletion potential induced between two spheres by saw polymers for a surprisingly wide range of @xmath126 . since
the semi - empirical potential ( [ eq3.21 ] ) reduces to the regular derjaguin form ( [ eq3.10 ] ) , if @xmath127 is replaced by @xmath42 , its scaling properties with density should be the same as those discussed in the previous subsections .
it is interesting to compare the results for interacting polymers with the results for non - interacting polymers .
firstly , just as was found for two walls , the range for the depletion potential is independent of density for non - interacting polymers .
this is true both for the simpler ao model@xcite , as well as for more sophisticated theories and computer simulations@xcite . for a given @xmath0 ,
using a theory based on non - interacting polymers will always overestimate the range compared to the interacting case . since for our accurate semi - empirical potential the range @xmath127
is related to @xmath42 by a density independent factor related to @xmath5 , the _ relative _ overestimate in the range for the spherical case should be close to that found when comparing a non - interacting theory to an interacting one for the range of the depletion potential between two plates , as done in fig . [
fig : range ] . for the case of two walls the absolute value of the well - depth at contact given by ideal polymers or the ao model was _ underestimated _ w.r.t . the interacting case . within the derjaguin picture of eq .
( [ eq3.2 ] ) , this implies that non - interacting polymer models will underestimate the force at contact @xmath128 for two spheres in a bath of interacting polymers by a similar factor to that shown for @xmath120 in fig .
[ fig : w02-w0semi ] . in the inset of fig .
[ fig : vderj ] we compare the scaled well - depth at contact @xmath129 for the derjaguin approximation with that of the ao model , given in eq .
( [ eqa.1 ] ) , where an effective range parameter @xmath130 , given by eq .
( [ eqa.4 ] ) in appendix a , was used to take into account the deformation of the polymers around the colloids . for a given @xmath2 , the well - depth
@xmath131 scales linearly with @xmath7 , which is now an _ overestimate _ compared to the result for interacting polymers .
the origin of this behavior can be easily understood : while the strength of the force is strongly underestimated by the ao model , the range is overestimated .
therefore the integral in eq .
( [ eq3.1 ] ) shows a compensation of errors .
this explains why , at low densities , the ao approximation well depth is very close to the interacting polymer result . for example for @xmath132
we find for @xmath43 that @xmath133 while for @xmath134 this ratio is still only @xmath135 . attempting a naive improvement by replacing @xmath136 in the ao expression by the true pressure of a polymer solution results in a worse approximation for all densities . when modeling polymer - colloid mixtures , the colloids are usually treated as single particles .
it thus seems natural to attempt the same for the polymers .
we have recently succeeded in modeling linear polymers as `` soft colloids''@xcite , where each polymer is replaced by a single particle centered on its cm .
different coils interact via an effective pair potential acting between their cm
. these effective interactions @xmath137 between the cm of the polymers were extracted from the radial distribution functions @xmath138 with an ornstein zernike inversion technique .
this procedure is justified by a theorem which states that for any given @xmath138 and @xmath7 there exists a single unique pair - potential which reproduces that radial distribution function@xcite .
our @xmath137 have a near gaussian shape , with an amplitude of order @xmath139 and a range of the order of the radius of gyration @xmath0 . for @xmath140 accurate analytic forms as a function of @xmath141 and @xmath29 are available@xcite .
our input @xmath138 s were generated by computer simulations of @xmath71 saw polymers on a cubic lattice , and so the density dependent effective potentials we derived will by definition reproduce the same cm pair structure as found in the underlying saw polymer system . when used in conjunction with the compressibility equation@xcite
, these potentials also provide a very good representation of the equation of state@xcite .
a similar coarse - graining procedure was followed to describe polymers near a flat wall@xcite , where even ideal polymers form a depletion layer .
if these were to be modeled as single particles , there would be no polymer - polymer interaction , but there would still be an effective polymer - wall interaction @xmath142 .
we used direct simulations of @xmath71 saw polymers to extract the density profile @xmath16 near a hard wall . the @xmath142 s which reproduce the density profile at each given bulk density @xmath7
were calculated using a wall - ornstein - zernike procedure@xcite .
again accurate analytic fits for densities @xmath143 are available@xcite . since our effective polymer - polymer potentials provide a very accurate representation of the pressure @xmath25 , while the polymer - wall interactions are constrained to reproduce the correct density profile @xmath16 , and therefore the correct adsorption @xmath26 , eq .
( [ eq2.2 ] ) implies that our soft - colloid approach will reproduce the correct wall - fluid surface tension @xmath13 .
this explains why our soft - colloid approach reproduces the correct value for @xmath144)@xcite , the depletion potential at contact between two walls , since eq .
( [ eq2.1 ] ) implies that this can be expressed completely in terms of @xmath13 .
similarly the slope of the potential is given by the osmotic pressure @xmath25 which is also accurately represented in the soft - colloid approach .
deviations w.r.t .
direct simulations were only observed for larger distances @xmath145 , where the soft - colloid approach produced a more pronounced repulsive bump in the depletion potentials.@xcite . to calculate the depletion potential between spheres , we need to first derive the effective sphere - polymer cm potential @xmath146 which would exactly reproduce the density profile @xmath147 around a single sphere of radius @xmath2 .
this can again be done using an ornstein - zernike ( oz ) technique to invert the density profiles@xcite .
these density profiles were calculated in paper i , where we found that the range of the profiles shrinks with increasing density .
the cm profiles can penetrate into the spheres because the polymers can deform around them , an effect which becomes more pronounced with decreasing sphere size ( this does nt happen in the monomer representation of course ) .
similarly , the effective potentials @xmath146 , derived with our inverse oz approach , become relatively softer with decreasing @xmath2 , as shown in fig .
[ fig : vrsphereall]@xcite . for a fixed @xmath5
the potentials become more repulsive for increasing density , just as was found for the polymer - wall case@xcite .
note that this behavior is the opposite to that of the density profiles .
again , because these potentials are constrained to give the correct density profiles and related adsorptions , this soft colloid picture should also correctly reproduce the surface tension @xmath83 for a single sphere immersed in a bath of interacting polymers .
the same should hold for the related one - body insertion free energies @xmath148 described in paper i. the simulation of the depletion potentials induced by the `` soft colloids '' proceeds in a similar manner to the simulations with saw polymers , described in section[sawdepletion ] .
however , the implementation is much simpler , because the interactions are smooth and both types of particles are in continuous space .
the simulations are also about two orders of magnitude faster , since each polymer is represented by only one particle , instead of the @xmath149 units per polymer used for the saw model .
these simulations are compared for @xmath75 in fig .
[ fig : vrad16-ejm - eff ] to the direct saw simulations .
the agreement is very good for @xmath150 and reasonable for @xmath78 .
however for the highest density , @xmath151 , the soft - colloid approach breaks down .
exactly the same trend was found for the other two size ratios ( not shown here ) . since the soft - colloid approach reproduces the correct one - body densities @xmath147 , and the correct one - body free energy @xmath148 of immersing a single sphere into a bath of polymers for all densities @xmath29 in the dilute and semi - dilute regimes , it is perhaps surprising that for the case of two spheres , it breaks down rather abruptly for higher densities .
to further investigate this issue , we plot the polymer cm density profiles for two spheres fixed at distance @xmath152 for @xmath75 and @xmath151 in fig . [
fig : dens - plot ] .
this corresponds to a distance where the saw @xmath153 has effectively gone to zero . at this short distance
the saw polymers clearly penetrate more easily in between the two colloids than the soft particles do , resulting to less attraction between the colloids .
a similar effect was found for polymers between two walls@xcite , but there the fact that sum - rules constrain the value of the potential at contact means that this difference manifested itself instead in an enhanced repulsive bump in the potential . in both cases ,
we attribute the error in the soft - colloid approach at strong confinement and higher densities to a breakdown in the `` potential superposition approximation''@xcite , i.e. the assumption that a polymer confined between two surfaces feels an interaction which is simply the sum of the two interactions @xmath146 with each separate surface . while this would be correct for simple fluids ,
it is not correct here , since the close presence of one wall affects the interaction of the polymers with the second wall .
this is essentially due to the deformability of the polymers , which is not correctly taken into account for strong confinement . 8 cm when we do a similar analysis of the densities as done in fig .
[ fig : dens - plot ] for @xmath154 however , the differences in the density profiles are very small , suggesting that the soft - particle picture is not yet breaking down .
coarse - graining polymers as soft colloids is expected to be most useful for @xmath14 , since once the polymers become much larger than the colloids , they can easily wrap around the colloids , an effect not easily treated in our coarse - graining scheme . since for @xmath14
, we have found that phase - separation sets in below @xmath155@xcite , the limit of physical stability is reached before we encounter problems with our coarse - graining scheme for polymer - colloid mixtures . to understand further the nature of the depletion potentials between the colloidal particles , mediated by the polymers in the `` soft colloid '' picture , we now turn to the superposition approximation , which expresses the two body depletion interactions in terms of one - body properties of a single colloid in a bath of soft polymers .
if one colloidal particle is fixed at the origin , and one is fixed at * r * , then one can define an inhomogeneous one - particle density of the soft colloids @xmath156 for that fixed configuration of the two colloidal particles .
the effective force induced between two colloidal hs by the soft colloids can then be written as@xcite : @xmath157 were @xmath146 is the interaction between the hs particles and the soft colloids .
this expression has an obvious intuitive interpretation .
the total force on a hs particle at @xmath158 is the average of the sum of its interactions with all the soft colloids .
if there are no other hs particles , then the equilibrium distribution will be spherically symmetric , and this force will be zero .
however , if there is another hs particle at @xmath159 , it will perturb the density distribution of the soft colloids , which will in turn result in a different total force acting on the particle at @xmath158 . by definition ,
this is the effective or depletion force @xmath160 acting on a particle at @xmath158 , induced by the presence of a particle at @xmath159 .
( [ eq4.1 ] ) is in principle exact , but requires knowledge of the soft colloid density around two colloidal hs . by approximating the full ( cylindrically symmetric ) one - body density @xmath156 by a superposition of the one - body density @xmath147 around an isolated single sphere@xcite : @xmath161 the effective depletion force ( [ eq4.1 ] ) can be entirely expressed in terms of the ( radially symmetric ) problem of a single colloidal sphere immersed in a polymer solution .
the results of such a calculation for @xmath153 are compared in fig .
[ fig : vrad16-eff - sup ] for @xmath75 . for all three densities
, the superposition approximation closely follows the direct simulations for the soft - colloids .
this implies that the full one - body density @xmath156 does not differ much from the simple superposition of eq .
( [ eq4.2 ] ) , even for the highest density considered , @xmath79 .
the reason for this is most likely that the effective polymer
polymer cm interaction @xmath162 is rather weak . in a recent study ,
a similar good performance for the superposition approximation was found for a low density hard - core depletant when the hs - depletant interaction was fairly long ranged@xcite .
good agreement was also found in a similar study of star - polymer colloid mixtures@xcite where accurate star - polymer
wall potentials@xcite are derived based on a coarse - graining approach@xcite . in contrast , if the density of a hard - core depletant increases , then the superposition approximation is known to break down for increasing packing fraction@xcite .
the fact that the superposition approximation works well for the soft colloids at the higher densities , also raises an interesting question as to why the soft - colloid picture itself breaks down at the higher densities . by construction ,
the one - body density profiles are identical for the full polymer case and the soft - colloid picture .
presumably the reason for the difference in the two - body depletion interactions can be traced either to a large deviation from superposition for the full polymer case , or else to a breakdown of the potential superposition approximation , as discussed in ref .
the influence of changing @xmath146 is illustrated in fig .
[ fig : vrad16-eff - sup ] , where we used the @xmath146 appropriate for @xmath154 to calculate the depletion potential with the density profiles for @xmath79 .
this @xmath146 is less repulsive than the true @xmath146 , and leads to an important difference in the depletion potential .
these depletion interactions are therefore quite sensitive to the exact form of @xmath146 . in conclusion
then , the superposition approximation works remarkably well for @xmath163 , the regime where the soft - colloid picture provides the most accurate representation of the true depletion potentials .
since the density profiles @xmath147 can be easily calculated by our wall - oz approach , this means that one only needs the polymer - colloid interaction @xmath146 and the polymer - polymer interaction @xmath162 as input to reliably calculate the full two - body depletion interactions from the superposition approximation .
a useful measure of the strength of an interaction is the second osmotic virial coefficient @xmath164 , since this is experimentally accessible through a measurement of the osmotic pressure at low densities@xcite .
such virial coefficients are very sensitive to the strength and nature of the depletion interaction@xcite .
a recent study@xcite has shown that for many simple systems consisting of a hs like repulsion with an additional attractive potential , the location of the liquid - gas or fluid - fluid critical point can be correlated with the point where the reduced virial coefficient @xmath165 , where @xmath166 is the virial coefficient of a hs system with radius @xmath2 . for our depletion systems ,
the reduced virial coefficient is given by @xmath167 -1\right).\ ] ] the direct simulation results depicted in fig [ fig : vrad16-ejm - newpot ] can be used to calculate @xmath168 , and the results are shown in fig [ fig : b2 ] for @xmath77 , @xmath154 , and for @xmath79 . as expected , for a given density @xmath29 , the @xmath168 become progressively more negative with decreasing @xmath5 .
although we do nt include any explicit error bars in our plots , these may be rather large for increasing attraction because they appear exponentially in eq .
( [ eq6.1 ] ) .
we found in the previous sections that a simple semi - empirical potential @xmath169 , given by eq .
( [ eq3.21 ] ) , provided a nearly quantitative description of the depletion potentials between two spheres .
it turns out that for this potential the virial coefficient can be integrated analytically : @xmath170 ( d + 4 r_c ) + \pi r_c w ( d + 2r_c)^3 r_c \right ) \nonumber \\ & + & \left .
3 \sqrt{2 d } ( d -\pi r_c w ( d + 2 r_c)^2 ) erfi\left[\sqrt{\frac{\pi r_c d w}{2 } } \right ] \right)\end{aligned}\ ] ] where @xmath171 and @xmath172 , i.e. the explicit density dependence has been suppressed for notational clarity .
the imaginary error function @xmath173 $ ] is defined as @xmath174/i$ ] ; its value is real .
as can be seen in fig .
[ fig : b2 ] , this expression compares well with the direct simulation results , demonstrating that the potential ( [ eq3.21 ] ) is a good approximation to the true depletion potential .
for this reason , we also plot the theoretical @xmath164 for two smaller @xmath5 s .
the density @xmath29 at which one would expect a ( metastable ) critical point and a related fluid - fluid de - mixing decreases with decreasing @xmath5 . to compare with experiments , one should keep in mind that our parameter @xmath29 would be equal to the density of a reservoir of polymers kept at the same chemical potential as a full polymer - colloid mixture@xcite .
if the pair - potentials are of sufficiently short range , the fluid - fluid transition becomes metastable w.r.t .
the fluid - solid transition of the larger particles@xcite , which widens to a much larger .
it has been recently found that this generic widening of the fluid - solid transition is primarily driven by the value of the potential at contact@xcite ; other details such as the shape and range of the attractive part of the potential are less important .
roughly speaking , the fluid - solid liquidus curve widens to about half the packing fraction of the hard - sphere freezing transition when @xmath175@xcite .
when combined with the criterion that @xmath176 at the critical density , this can be used for a rough prediction of the @xmath5 below which the fluid - fluid transition is metastable w.r.t .
the fluid - solid one .
the minimum @xmath5 for which the fluid - fluid transition is still stable is between @xmath177 and @xmath178 , which is consistent with other calculations@xcite , and experiments@xcite . of course one
should keep in mind that the larger the size ratio @xmath5 , the more important three - body and higher order interactions become@xcite .
these pair - potential criteria should therefore become less accurate for increasing @xmath5 .
both rg theory@xcite and integral equation approaches@xcite predict that for large enough @xmath5 , @xmath168 should go through a minimum as a function of @xmath29 , although the two theories differ significantly on quantitative details .
experiments seem to show similar effects@xcite . remarkably , expression ( [ eq6.2 ] ) , which is based on the simple potential of eq .
( [ eq3.21 ] ) also shows a minimum , as shown in fig .
[ fig : b2-smallq ] . even more surprisingly , considering how the potential was derived , is that the minimum is near @xmath43 , as predicted by rg theory , and that it closely follows the asymptotic law @xmath179 , derived by eisenriegler@xcite ( see the inset of fig .
[ fig : b2-smallq ] ) .
in contrast , when the simpler derjaguin expression ( [ eq3.10 ] ) is used for @xmath168 , no minimum as a function of @xmath29 appears for large @xmath5 . at present
it is not clear why our expression for @xmath168 , eq .
( [ eq6.2 ] ) , should agree so well with the rg results in this large @xmath180 regime .
we have used a combination of computer simulations and theory to derive the depletion interactions induced by excluded volume polymers .
our main computer simulation results are depicted in fig .
[ fig : forces ] and fig .
[ fig : vrad16-ejm - newpot ] and our main theoretical results are the depletion potential for two plates , given by eq .
( [ eq2.6 ] ) and the depletion potential for two spheres , given by eq .
( [ eq3.21 ] ) . for two plates
, we found that a simple linear depletion potential ( [ eq2.6 ] ) with a well - depth of @xmath181 , and a slope of @xmath25 is very accurate .
the range decreases with density , with the largest relative change occurring in the dilute regime . in the semi - dilute regime ,
the well - depth scales as @xmath182 , while the range scales as @xmath183 .
the depletion potential also takes on a simpler functional form , given by eq .
( [ eq2.7 ] ) .
these results differ significantly from theories based on ideal polymers . for two spheres , we find that the simple derjaguin approximation works much better than expected due to a cancellation of errors related to the deformation of the polymers around a sphere .
interestingly , the derjaguin expression for the force at contact seems particularly accurate , even for very large size ratios @xmath5 .
we used this observation , together with a correction for the derjaguin range , to derive the depletion potential @xmath184 ( eq . ( [ eq3.21 ] ) ) , which was shown to agree quantitatively with our direct computer simulation results .
while the excellent agreement with simulations is gratifying , further study is needed to understand why our main ansatz , that the force at contact is accurately approximated by the derjaguin approximation , works so well for such a wide range of @xmath5 . the range of our potential scales with density in the same way that the range for the two plates does .
but in contrast to the case of two plates , where the well depth at contact is more attractive than that for ideal polymers , the well depth at contact for two spheres is less attractive than that predicted by ideal polymer theories .
for example , in the semi - dilute regime , @xmath185 , compared to the scaling of the ao potential : @xmath186 . in other words ,
ideal polymer theories overestimate both the range _ and _ the well - depth of the depletion potential between two spheres , induced by interacting polymers .
we also calculated the depletion potential between spheres induced by polymers coarse - grained as `` soft colloids '' .
for the dilute regime , good agreement with the direct simulations of saw polymers was achieved . however , for the highest density , @xmath151 , the polymers as soft colloids approach strongly overestimates the depth of the depletion potential .
we traced this discrepancy to a breakdown of the potential superposition approximation , something also found for two plates@xcite .
however , since our coarse - graining scheme is mainly useful for @xmath14 , where phase - separation sets in for @xmath163 , this breakdown at high density is not relevant to the equilibrium phase behavior of colloid - polymer systems .
in fact , we have recently used this scheme to calculate the phase - separation binodals of polymer - colloid mixtures for several size ratio s @xmath187@xcite .
direct simulations of saw polymers and colloids would be significantly more expensive .
the advantage over directly using a pair - potential such as eq .
( [ eq3.21 ] ) , is that many - body interactions are also effectively taken into account@xcite .
virial coefficients can provide a sensitive measure of the quality of an effective potential .
we were able to integrate eq .
( [ eq3.21 ] ) to derive an analytical representation of the virial coefficient ( [ eq6.2 ] ) , which compares very well with the direct simulation results .
surprisingly , this same analytic form agrees quantitatively with rg results in the limit @xmath188 .
further study is needed to clarify why eq .
( [ eq6.2 ] ) works so well in this limit .
a careful comparison with experiments is complicated by the fact that most measurements are of phase behavior or structure , which depend only indirectly on the depletion interactions .
the most extensive direct measurements of the depletion potential between spheres , carried out by verma _
et al._@xcite , were on a system with semi - flexible polymers ( dna ) which do not follow the same scaling laws as the saw polymers we treated in this paper .
however , given a good model for the equation of state and the adsorption @xmath26 for these polymers , it would be straightforward to calculate the depletion potentials using methods derived in this paper
. a detailed comparison with experiments will be the subject of an upcoming review@xcite .
this paper has concerned itself with the case of interacting , non - adsorbing polymers in a good solvent . by taking eqs .
( [ eq2.1 ] ) , ( [ eq2.2 ] ) , ( [ eq2.6 ] ) , and ( [ eq3.21 ] ) together , if follows that accurate depletion potentials can be calculated with only a knowledge of the adsorption @xmath26 and the equation of state @xmath25 .
this implies that the effect of changing solvent quality , or of adding attractions or repulsions between the colloids and the polymers can be understood by seeing how they effect these two quantities .
note that this is different from binary colloid mixtures , where the effects of added interparticle attractions and repulsions have a more subtle effect on the resultant depletion potentials@xcite .
it is not hard to see that for polymers in a good solvent , adding an attractive ( repulsive ) polymer - colloid interaction results in a less ( more ) attractive depletion potential , and possibly even to repulsive effective interactions .
this follows since @xmath25 is unchanged , and only @xmath26 is affected in an obvious way .
an example where this may be relevant concerns the common practice to stabilize colloidal particles by a short polymer brush .
the interaction of the free polymers with the polymer brush can be quite subtle@xcite , but our theory implies that it is only the overall change in the adsorption which effects the depletion potential , and by extension , the phase behavior .
changing the solvent quality has a less obvious effect , since both @xmath25 and @xmath189 are affected . for a given density
, we expect the absolute magnitudes of both to decrease , so that the strength of the depletion potential should decrease as well .
of course , for strong deviations from the case of non - adsorbing polymers in a good solvent , some of the simplifying assumptions that went in to the derivation of the depletion potentials may begin to break down .
for example , for very strong adsorption , the system may show bridging flocculation@xcite , or gelation@xcite .
a rich phenomenology may be expected ; this will be the subject of an upcoming investigation@xcite .
aal acknowledges support from the isaac newton trust , cambridge , and the hospitality of lydric bocquet at the ecole normale superieure in lyon , where much of this work was carried out .
ejm acknowledges support from the royal netherlands academy of arts and sciences . and support from the stichting nationale computerfaciliteiten ( ncf ) and nederlandse organisatie voor wetenchappelijk onderzoek ( nwo ) for the use of supercomputer facilities .
we thank r. evans and h. lwen for helpful discussions .
the asakura oosawa model , where the ideal polymers are modeled as interpenetrable spheres of radius @xmath108 , was first introduced in 1958@xcite . between two plates
this results in the linear depletion potential similar to eq .
( [ eq2.10 ] ) , with a range @xmath108 , while between two spheres it results in the well - known asakura - oosawa form @xmath190 which is exact within the confines of the ao model .
here we defined @xmath191 .
similarly , the use of the derjaguin approximation ( [ eq3.2 ] ) together with the linear ao potential between two plates results in : @xmath192 comparison of the two expressions at contact , where @xmath193 and @xmath194 , shows that the derjaguin expression ( [ a.2 ] ) underestimates the well depth at contact w.r.t . the exact expression([eqa.1 ] ) , a discrepancy which gets relatively worse for decreasing @xmath195 .
this is illustrated in fig [ fig : vao - derj](a ) for the same size ratios we used for the interacting polymers . since the derjaguin approximation is only valid for large @xmath195
, this breakdown is expected . before using the ao model to describe polymers , one needs to fix the parameter @xmath108 .
a common prescription has been to take @xmath197@xcite . however , between two walls , one should take @xmath198 , which corresponds to approximating the depletion layer of ideal polymers by a step - function with the same depletion volume .
this then results in a well - depth which is equal to that found for ideal polymers@xcite .
however , for polymers of a given size @xmath0 , the width of the depletion layer around a single sphere will decrease with decreasing @xmath2 , because the polymers can partially wrap around the colloidal particles@xcite . by equating the one - body insertion free energy for a hs sphere into a bath of ao model particles to that found for inserting a hs into a bath of ideal polymers
, we found , in paper i , an exact analytic prescription for the effective parameter @xmath130 @xmath199 which decreases with decreasing @xmath2 as expected .
just because the one - body terms are equal does not mean that the same prescription for @xmath108 will work for the two - body terms . but direct computer simulations@xcite have shown that using @xmath130 in the full ao expression ( [ eqa.1 ] ) results in a much better approximation of the depletion potential between two spheres in a bath of ideal polymers , than using a prescription which fixes @xmath108 independently of @xmath2 . with this in mind , we now compare again the derjaguin expression ( [ a.2 ] ) with @xmath200 to the full ao expression ( [ eqa.1 ] ) but now with an effective ao radius @xmath130 from eq .
( [ eqa.4 ] ) appropriate to the desired @xmath2 . as demonstrated in fig .
[ fig : vao - derj](b ) , the derjaguin approach is now a more faithful approximation of the full ao model depletion potential , which is itself a better approximation of the true depletion potential induced by ideal polymers .
although the derjaguin approach does nt capture the slight reduction in range , the well - depth is much closer to the ao value .
of course this improvement arises from a cancellation of errors .
however , for the interacting polymers the depletion potential between two walls is also well described by a linear form , and we expect the same deformation effect to be relevant for polymers near spheres
. therefore , we might hope for a similar cancellation of errors when the derjaguin expression is used for interacting polymers .
likos , phys . rep . *
348 * , 267 ( 2001 ) .
louis , phil .
a * 359 * , 939 ( 2001 ) .
hansen and i.r .
mcdonald , _ theory of simple liquids , 2nd ed . _
( academic press , london , 1986 ) .
d. frenkel and b. smit , _ understanding molecular simulations _
( academic press , 1995 ) .
s. asakura and f. oosawa , j. chem . phys .
* 22 * , 1255 ( 1954 ) . s. asakura and f. oosawa , j. polym .
, polym . symp . * 33 * , 183 ( 1958 ) , a. vrij , pure appl . chem . *
48 * , 471 ( 1976 ) .
a. p. gast , c.k .
hall , and w.b .
russel , j. colloid interface sci . * 96 * , 251 ( 1983 ) .
lekkerkerker , w.c.k .
poon , p.n .
pusey , a. stroobants and p.b .
warren , europhys .
lett . * 20 * , 559 ( 1992 ) .
f.l . calderon , j. bibette , and j. bais , europhys .
* 23 * , 653 ( 1993 ) .
ilett , a. orrock , w.c.k .
poon , and p.n .
pusey , phys .
e , * 51 * , 1344 ( 1995 ) .
a. weiss , k. d. hrner , and m. ballauff , j. colloid interface sci .
* 213 * , 417 ( 1999 ) .
a. moussaid , w.c.k .
poon , p.n .
pusey , and m.f .
soliva , phys .
lett . * 82 * , 225 ( 1999 ) .
a. a. louis , r. finken , and j.p .
hansen , europhys .
lett . * 46 * , 741 ( 1999 ) m. dijkstra , j. brader and r. evans , j. phys . : condens .
matter * 11 * , 10079 ( 1999 ) .
et al . _
rev . lett . * 78 * , 3963 ( 1997 ) .
r. verma , j.c .
crocker , t.c .
lubensky , and a.g .
yodh , phys .
lett . * 81 * , 4004 ( 1998 ) ; macromolecules * 33 * , 177 ( 2000 ) . c. bechinger , d. rudhardt , p. leiderer , r. roth and s. dietrich , phys . rev
* 83 * , 3960 ( 1999 ) .
j. f. joanny , l. leibler , and p.g . de gennes , j. polym .
* 17 * , 1073 ( 1979 ) .
fleer , j.m.h.m .
scheutjens , and b. vincent , acs symp .
* 240 * , 245 ( 1984 ) .
walz and a. sharma j. colloid interface sci . * 168 * , 485 ( 1994 ) .
warren , s.m .
ilet , and w.c.k .
poon , phys .
e * 52 * , 5205 ( 1995 ) .
r. dickman and a. yethiraj , j. chem . phys . *
100 * , 4683 ( 1994 ) .
a. hanke , e. eisenriegler and s. dietrich , phys .
e. * 59 * , 6853 ( 1999 ) . .
f. schelesenger , a. hanke , r. klimpel , and s. dietrich , phys .
e * 63 * , 041803 ( 2001 ) .
r. tuinier and h.n.w .
lekkerkerker , eur .
j. e * 6 * , 129 ( 2001 ) .
r. roth , r. evans , and s. dietrich , phys . rev .
e * 62 * , 5360 ( 2000 ) .
b. gtzelmann , r. roth , s. dietrich , m. dijkstra , and r. evans , europhys .
lett . * 47 * , 398 ( 1999 ) ; r. roth , doctoral thesis , bergische universitt wuppertal ( 1999 ) .
louis and r. roth j. phys . :
matter*33 * , l777 ( 2001 ) ; r. roth , r. evans , and a.a .
louis , phys .
e. * 64 * , 051202 ( 2001 ) .
m. dijkstra , r. van roij and r. evans , phys .
81 * , 2268 ( 1998 ) ; _ ibid _ * 82 * , 117 ( 1999 ) ; phys .
e. * 59 * , 5744 ( 1999 ) .
meijer and d. frenkel , phys .
lett . * 67 * , 1110 ( 1991 ) ; j. chem . phys . * 100 * , 6873 ( 1994 ) . m. fuchs and k.s .
schweizer , phys .
e * 64 * 021514 ( 2001 ) .
de gennes , c.r .
b * 288 * , 359 ( 1979 ) . p.g de gennes , _ scaling concepts in polymer physics _ , ( cornell university press , ithaca ) .
e. eisenriegler , j. chem . phys . * 113 * , 5091 ( 2000 ) .
louis , p.g .
bolhuis , j.p .
hansen and e.j .
meijer , cond - mat/0111518 m. doi and s.f .
edwards , _ the theory of polymer dynamics _ , ( oxford university press , oxford , 1986 ) .
derjaguin , kolloid zeits . * 69 * , 155 ( 1934 ) .
louis , p.g .
bolhuis , j.p .
hansen and e.j .
meijer , phys .
85 * , 2522 ( 2000 ) .
bolhuis , a.a .
louis , j.p .
hansen , and e.j .
meijer , j. chem . phys . * 114 * , 4296 ( 2001 )
bolhuis , a.a .
louis , and j.p .
hansen , phys .
e. * 64 * , 021801 ( 2001 ) .
bolhuis , a.a .
louis , macromolecules * 35 * , 1860 ( 2002 ) .
it is generally most natural to analyze depletion systems in a semi - grand ensemble , where the colloidal particles are in osmotic equilibrium with a reservoir of polymers at fixed chemical potential @xmath18@xcite . throughout this paper , therefore , the polymer density @xmath7 is actually that of a reservoir of polymers at the same chemical potential @xmath18 .
for the two - particle depletion problem in the thermodynamic limit , this is equivalent to the actual polymer density , but if one wants to use the depletion potential to analyze the phase behavior of a many colloid system , this prescription should be kept in mind .
y. mao , p. bladon , h.n.w .
lekkerkerker and m.e .
cates , mol .
phys . * 92 * , 151 ( 1997 ) r. maasen , e. eisenriegler , and a. bringer , j. chem . phys . *
115 * , 5292 ( 2001 )
. it would be straightforward to generalize the potential to one with a slightly more rounded shape , such as that found for ideal polymers@xcite .
although this would be slightly more accurate , the differences are so small that we will stick to the simpler linear form for this paper . the definition for the range would remain the same for both versions of the potential .
noro and d. frenkel , j. chem . phys . *
113 * , 2941 ( 2000 ) .
j. van der gucht , n. a. m. besseling , j. van male and m. a. cohen stuart , j. chem . phys . * 113 * , 2886 ( 2000 ) . y. mao , m. e. cates , and h. n. w. lekkerkerker , physica a * 222 * , 10 ( 1995 ) .
louis , e. allahyarov , h. lwen and r.roth condmat/0110385 .
a. broukhno , b. jonsson , t. akesson , and p.n .
vorontsov - velyaminov , j. chem . phys . * 113 * , 5493 ( 2000 ) .
b. vincent , j. edwards , s. emmett , and r. croot , colloids and surfaces * 31 * , 267 ( 1988 ) .
r. tuinier and h.n.w .
lekkerkerker , preprint ( 2001 ) . to compare to the results of @xcite we used their expression for the end - to - end distance @xmath201 for the three dimensional fluctuating bond model , to derive the radius of gyration at zero density through @xmath202@xcite .
the two results with the highest monomer density are not shown since they appear to be closer to the melt than to the semi - dilute regime .
m. triantafillou and r.d .
kamien , phys .
e * 59 * , 5621 ( 1999 ) .
r. tuinier , g. a. vliegenthart , and h. n. w. lekkerkerker , j. chem . phys . * 113 * , 10768 ( 2000 ) .
louis , p.g .
bolhuis , r. finken , v. krakoviack , e.j .
meijer and j.p .
hansen , to be published in physica a , condmat/011209 r.l .
henderson , phys .
a. * 49 * , 197 ( 1974 ) ; j.t .
chayes and l. chayes , j. stat .
* 36 * , 471 ( 1984 ) .
l. reatto , phil . mag .
a * 58 * , 37 ( 1986 ) ; l. reatto , d. levesque , and j.j .
weis , phys .
rev . a. * 33 * , 3451 ( 1986 ) .
p. attard , j. chem .
* 91 * , 3083 ( 1989 ) .
t. biben , p. bladon , and d. frenkel , j. phys .
condensed matter * 8 * , 10799 ( 1996 ) .
j. dzubiella and c.n .
likos , preprint , to appear in europhys .
a. jusufi , j. dzubiella , c. n. likos , c. von ferber , h. lwen , j. phys . :
condensed matter * 13 * 6177 - 6194 ( 2001 ) .
likos h. lwen , m. watzlawek , b. abbas , o. jucknischke , j. allgaier , and d. richter , phys .
* 80 * 4450 ( 1998 ) .
m. watzlawek , c.n .
likos , and h. lwen , phys .
lett . * 82 * , 5289 ( 1999 ) .
s. amokrane , j. chem . phys . * 108 * , 7459 ( 1998 ) . h. de hek and a. vrij , j. coll .
sci . * 88 * , 258 ( 1982 ) .
a. m. kulkarni , a. p. chatterjee , k. s. schweizer , and c. f. zukoski , phys .
lett . * 83 * , 4554 ( 1999 ) .
g. a. vliegenthart and h. n. w. lekkerkerker , j. chem . phys . * 112 * , 5364 ( 2000 ) . m. h. hagen and d. frenkel , j. chem . phys .
* 101 * 4093 ( 1994 ) .
louis , r. finken and j.p .
hansen , phys .
e. * 61 * , r1028 ( 2000 ) .
d. goulding and s. melchionna , phys .
e * 64 * 011403 ( 2001 ) .
a.p . chatterjee and k.s .
schweizer , j. chem .
phys , * 109 * , 10471 ( 1998 ) .
bolhuis , a.a .
louis , e.j .
meijer and j.p .
hansen , manuscript in preparation .
louis , in preparation for j. phys . : condens . matter ( 2002 ) .
a. jones and b. vincent , colloids and surfaces , * 42 * 113 ( 1989 ) .
a. johner , j.f .
joanny , s. diez orrite , and j. bonet avalos , europhys lett . * 56 * , 549 ( 2001 ) .
v. krakoviack , a.a .
louis , and j.p .
hansen , work in preparation .
e. eisenriegler , a. hanke , and s. dietrich , phys .
e * 54 * , 1134 ( 1996 ) . | the depletion interactions between two colloidal plates or between two colloidal spheres , induced by interacting polymers in a good solvent , are calculated theoretically and by computer simulations .
a simple analytical theory is shown to be quantitatively accurate for case of two plates .
a related depletion potential is derived for two spheres ; it also agrees very well with direct computer simulations .
theories based on ideal polymers show important deviations with increasing polymer concentration : they overestimate the range of the depletion potential between two plates or two spheres at all densities , with the largest relative change occurring in the dilute regime .
they underestimate the well depth at contact for the case of two plates , but overestimate it for two spheres .
depletion potentials are also calculated using a coarse graining approach which represents the polymers as `` soft colloids '' : good agreement is found in the dilute regime .
finally , the effect of the polymers on colloid - colloid osmotic virial coefficients is related to phase behavior of polymer - colloid mixtures . 2 |
the current standard cosmological scenario , supported by observations of the cosmic microwave background ( cmb ) and of the large - scale structure ( lss ) , appears to be consistent with the @xmath4cdm concordance model , where the universe is dominated by cold dark matter ( cdm ) and its accelerating expansion driven by a cosmological constant @xmath4 or dark energy ( de ) .
a recent strong support of this paradigm has been presented by park et al .
( 2012 ) , who was able to prove that observed high- and low - density lsss have the richness / volume and size distributions consistent with the @xmath4cdm universe .
in addition , the primordial density perturbations from which halos and galaxies form are assumed to be a gaussian random field , as predicted by inflationary theories ( guth 1981 ; linde 1982 ; bardeen et al .
1986 ) ; state - of - the - art data from the wilkinson microwave anisotropy probe ( wmap ; spergel et al .
2003 , komatsu et al .
2011 ) , the sloan digital sky survey ( sdss ; york et al .
2000 ; stoughton et al . 2002 ; abazajian et al .
2009 ) or the wigglez survey ( blake et al . 2011 ) are still favoring this hypothesis .
however , some claims or hints of primordial non - gaussianity have recently appeared in the literature ( jeong & smoot 2007 ; yadav & wandelt 2008 ; komatsu et al .
2009 , 2011 ; slosar et al . 2008 ; smidt et al . 2010 ) , and challenged the validity of the simplest inflationary paradigm .
indeed , if detected , primordial non - gaussianity would indicate a structure formation scenario different from the concordance cosmological model , and force us to revise the physics of the very early universe along with several aspects of the lss dynamics ( but see also hwang 2012 for a more general discussion on modern cosmology ) . to this end , topology - related statistics offer a precious benchmark for testing the underlying gaussianity of the initial density field , since topology can be regarded as an important physical property of the matter density that can be compared with predictions of the simplest inflationary models , where gaussian random phase initial conditions are generated from quantum fluctuations of an inflaton field in the early universe .
in addition , topology measured at the present epoch should reflect that of the initial conditions on smoothing scales considerably larger than the correlation length , because fluctuations which are still in linear regime maintain their initial topology ( see gott et al .
1987 , who confirmed this property with @xmath0-body simulations ) ; this fact allows one to test directly the gaussian paradigm , and permits to use topology as a cosmic standard ruler ( park & kim 2010 ) . from the theoretical side , since the pioneering work of gott et al .
( 1986 ) , a variety of analytic and numerical tools to analyze observational and simulated data for measuring topology have long been developed mainly using the genus statistics to quantify the topology of isodensity contours ( gott et al .
1987 , 1989 ; hamilton et al . 1986 ; vogeley et al . 1994 ; park et al .
2005a , 2005b ) .
in particular , the analytic prediction for the genus curve of a gaussian field in linear regime is well - known ( gott et al .
1986 ) , and its perturbative expression in the weakly nonlinear regime has also been obtained ( matsubara 1994 ) ; this lognormal model turned out to be a good empirical approximation in the strongly nonlinear regime ( hikage et al .
2002 ) . along with analytic tools ,
large - volume @xmath0-body simulations are routinely used to quantify several systematics which affect the genus curve , such as finite pixel size , sparse sampling , peculiar velocity distortions in redshift space or survey boundaries .
the ability to correct for these effects is essential , as the remaining small deviations from the random phase curve give important information about the physics connected with galaxy formation , nonlinear gravitational clustering , and primordial non - gaussianity if any .
in fact , on smaller scales nonlinear gravitational evolution and biased galaxy formation make the topology of the observed galaxy distribution deviate from the gaussian form , even if the initial conditions were gaussian - distributed . using fractional volume rather than direct density threshold as the independent variable in topology analysis mitigates but does not fully eliminate these nonlinear and biasing effects ( weinberg et al .
1987 ; melott et al .
ultimately , all the secondary non - gaussianities need to be disentangled from the primordial contribution , and this can now be done very accurately , without assuming any ` a priori ' model for the underlying signal . from the observational side ,
a long list of studies have been pursued on a variety of datasets ( see for example park , gott , & da costa 1992 ; park et al .
1992 , 2005b ; park , gott , & choi 2001 ; moore et al .
1992 ; vogeley et al . 1994 ; rhoads et al . 1994 ; protogeros & weinberg 1997 ; canavezes et al . 1998 ; hoyle et al .
2002 ; hikage et al .
2002 , 2003 ; james et al .
2007 , 2009 ; gott et al 2008 , 2009 ; choi et al .
they focused on characterizing the three - dimensional topology , and showed that , depending on the considered scale , topology is useful in constraining both cosmological parameters and the galaxy formation mechanism .
for example , park et al .
( 2005b ) characterized the topology of the sdss main galaxy sample from the nyu value added galaxy catalog ( nyu vagc ; blanton et al .
2005 ) , which has similar sky coverage as the sdss data release 4 ( dr4 ; adelman - mccarthy et al .
2006 ) , and presented the first clear demonstration of luminosity dependence of galaxy clustering topology ( i.e. brighter galaxies show a stronger signal of meatball topology ) ; more recently , gott et al .
( 2009 ) measured the three - dimensional lss topology of the sdss dr4plus lrg sample from the nyu vagc ( a subsample of the sdss dr5 ; adelman - mccarthy et al .
2007 ) and found strong consistency with gaussianity of the primordial fluctuations . in the latter case ,
the large sample size available allowed topology to be an important tool for testing galaxy formation models .
also , choi et al . (
2010 ) measured the topology of the main galaxy distribution using the sdss dr7 ( abazajian et al . 2009
; choi , han , & kim 2010 ) , studied the scale - dependent topology bias , and examined the dependence of galaxy clustering topology on galaxy properties ( i.e. luminosity , morphology , color ) at different smoothing scales .
the large volume - limited sample enabled an unprecedented measurement of the genus curve , with an amplitude of @xmath5 at 6 @xmath3mpc smoothing scale and an estimated uncertainty of 4.8@xmath6 , including all systematics and cosmic variance .
in addition , choi et al . (
2010 ) detected deviations of the genus curve from gaussianity , interpreted as the fact that voids and superclusters are more connected and their sizes are larger than those expected for gaussian fields .
the same authors then used these results to test five different galaxy formation models , which indeed are tuned to reproduce the two - point correlation function and the luminosity function but not high - order statistics , and found significant discrepancies : none of the models could reproduce all the aspects of the observed clustering topology . while the significant discrepancies in the sdss dr7 galaxy clustering topology at nonlinear scales found by choi et al .
( 2010 ) are mainly driven by the inaccuracy of galaxy formation models , a cleaner problem is to consider the topology of lrgs instead which is the main focus of this paper .
this is because the lrg sample covers a much larger and deeper volume ( i.e. it allows one to observe topology at the largest scales ) , essentially in the linear regime .
in addition , lrgs are particularly useful in refining cosmological parameters ( tegmark et al . 2006 ) , and are expected to play an important role in characterizing de through the ratio of de pressure to energy density ( see bassett et al .
2005 ; eisenstein et al . 2005 ; percival et al .
a number of previous analyses have addressed the clustering of lrgs , especially in relation to baryon acoustic oscillation ( bao ) science ( see for instance ross et al . 2012 , and references therein ) . on the contrary ,
fewer studies have been based on topology . among those ,
gott et al . ( 2009 ) measured the three - dimensional genus topology of lrgs , using two volume - limited samples constructed from the sdss dr4plus sample , i.e. a dense shallow sample with @xmath7 smoothing , and a sparse deep sample with @xmath8 smoothing .
the amplitude of the genus curve was found to reach about 167 with a @xmath9 uncertainty at @xmath7 scale .
a major result of their study was that topology of lrgs in the sdss agrees very well with that of mock galaxies in the @xmath4cdm universe with the same cosmological parameters : small distortions in the genus curve , expected from nonlinear biasing and gravitational effects , are well explained by @xmath0-body simulations with a subhalo finding technique adopted to locate lrgs .
this suggests that the formation of lrgs can be modeled well without any free - fitting parameters .
the main goal of this paper is to characterize the three - dimensional genus topology of spectroscopic lrgs using the sdss dr7 catalog , improving on the previous results presented by gott et al .
in particular , we strive to carefully model and remove all known systematics which affect the observed genus ( i.e. finite pixel size , survey boundary , radial and angular selection function and shot noise ) , and estimate the uncertainties in the measured genus accurately .
this is achieved by comparing our measurements with 81 mock sdss lrg surveys along the past light cone constructed from the horizon run 3 ( hr3 ; kim et al .
2011 ) , one of the largest @xmath0-body simulations to date that evolved @xmath1 particles in a @xmath2mpc size box .
our main result for the observed genus curve is shown in figure [ fig : main_result ] ( red solid line ) , and compared with a previous topology measurement on the sdss dr4plus dataset ( blue solid line ; gott et al . 2009 ) .
the main point of the figure is to show the dramatic increase in amplitude of the genus curve over the two different datasets ( i.e. the sdss dr4plus versus the sdss dr7 ) in the same redshift range from 0.16 to 0.36 and rest - frame @xmath10-band absolute magnitudes of @xmath11 , due to the much larger volume that is covered .
in fact , we find the genus amplitude to reach 285 with an uncertainty of 4.0% at @xmath12 gaussian smoothing scale including cosmic variance ( the most accurate measurement to date ) , while gott et al .
( 2009 ) found the genus curve to reach about 167 with a @xmath9 uncertainty at @xmath7 smoothing scale ; for comparison with a different galaxy population , park et al .
( 2005b ) and choi et al .
( 2010 ) reported an uncertainty of @xmath13 and @xmath14 at 5@xmath3mpc and 6@xmath3mpc scales , respectively , for the genus obtained from the main galaxy sample of the sdss dr3 and dr7 .
we will return to this measurement in great detail in section [ sec : genus_top ] , while in a forthcoming publication we interpret the deviation of the genus curve from the expected gaussian prediction in the context of primordial non - gaussianity .
the layout of the paper is organized as follows . in section [ sec : theory ] , we present the main theoretical framework of this study ; in particular , we review the genus statistics for gaussian fields and discuss how to extend the formalism for non - gaussian fields . in section [ sec : data_methods ] , we describe the sdss lrg sample used for our measurements and the methodology applied to the observational data . in section
[ sec : sims_methods ] , we present the hr3 @xmath0-body simulation and explain the procedure adopted to construct the 81 mock lrg surveys . in section [ sec : genus_top ] , we show our results for the lrg genus statistics , compare measurements from sdss dr7 data and simulations and quantify the non - gaussian deviations of the genus curve with genus - related quantities . in section [ sec : genus_sys ] , we discuss the effects of known systematics on the genus , and present the genus curve after corrections for systematics . we conclude in section [ sec : summary ] , and leave some more details on the genus curves in the appendix .
we begin by revisiting the basic theory of the genus for gaussian fields and by introducing some genus - related statistics .
we also briefly review the formalism for describing the non - gaussian effect on the genus curve in the weakly nonlinear regime according to second - order perturbation theory , originally derived by matsubara ( 1994 , 2003 ) .
we will later compare the theory outlined here with results from the sdss dr7 lrg dataset , and with measurements from simulated lrg sample .
the full extension to non - gaussian fields with the inclusion of primordial non - gaussianity will be presented and discussed in a forthcoming publication .
the genus is a measurement of the topology of isodensity contour surfaces in a smoothed galaxy field ( gott et al .
1987 ) . in mathematical terms
, it is defined as follows .
consider a three - dimensional gaussian random field @xmath15 with @xmath16 the spatial coordinate , and measure the topology of the excursion regions where @xmath17 is equal to , or is above a given threshold level @xmath18 . here
@xmath19 is the mean of the field @xmath17 , @xmath20 its root mean square ( @xmath21 ) value , and @xmath22 .
denote with @xmath23 the space which contains the set of the excursion regions ( i.e. a 3-manifold subset ) , and indicate its boundaries with @xmath24 ( i.e. a 2-manifold subset ) .
for each component @xmath25 of @xmath24 , according to the gauss - bonnet theorem , the mathematical genus satisfies the following relation @xmath26 where @xmath27 is the euler characteristic of the surface of the three - dimensional excursion region ( i.e. the integrated gaussian curvature of the surface ) .
hence , the total genus of the boundary @xmath24 becomes @xmath28 where @xmath0 is the number of components of @xmath24 see park et al .
( 2013 ) for a full derivation of the previous formula .
in cosmology , the standard definition of genus slightly differs from the previous mathematical one , since the genus is defined as the number of holes minus the number of isolated regions in the isodensity contour surfaces , at a given threshold level @xmath29 .
namely , @xmath30 the relation between the two definitions is simply expressed by @xmath31 , where @xmath0 has been defined above . for further insights on topological invariants , and for the mathematical connection between the cosmological genus and the betti numbers for excursion sets of gaussian random fields , we refer the reader to park et al .
( 2013 ) . in the case of gaussian fields , the genus per unit volume @xmath32 as a function of density threshold level @xmath29 is known ( i.e. doroshkevich 1970 ; adler 1981 ; hamilton , gott , & weinberg 1986 ) : @xmath33 the amplitude @xmath34 is given by @xmath35 while the spectral moments of the fields , @xmath36 , are computed from @xmath37 in the previous relation , @xmath38 is the power spectrum smoothed on a scale @xmath39 by a window function @xmath40 , where @xmath41 and @xmath42 is the matter power spectrum . in particular , in this study we adopt a gaussian smoothing of the form @xmath43 .
note also that , for @xmath44 , @xmath45 is the variance of the fluctuating field , while @xmath45 is the variance of its derivative when @xmath46 . to separate the variation in topology from the change of the one - point density distribution , in this work we also measure the genus as a function of the volume - fraction threshold @xmath47 ( as opposed to the direct density threshold @xmath29 )
this parameter defines the density contour surface such that the volume fraction @xmath48 in the high density region is the same as the volume fraction in a gaussian random field contour surface having @xmath49 , namely : @xmath50 after correcting for known systematics , deviations of the observed genus curve from the gaussian expectation ( i.e. eq .
[ eq_genus_standard ] ) are due to nonlinear gravitational evolution and non - gaussianity of the primordial density field .
a number of studies in the literature have already addressed the impact of non - gaussianity on the genus curve ( see for example weinberg et al .
1987 ; park & gott 1991 ; park , kim , & gott 2005a ) . in
what follows , we briefly discuss the non - gaussian effect on the genus curve caused by nonlinear gravitational evolution in the weakly nonlinear regime ( which tends to distort the gaussian expectation for the genus statistic ) , in the context of second - order perturbation theory along the lines of matsubara ( 1994 , 2003 ) .
more details on the non - gaussian modifications of the genus curve will be presented in young - rae kim et al .
( in preparation ) . to first order in @xmath20 , the nonlinear correction that one must apply to the genus curve due to gravitational evolution is an odd function of the threshold @xmath29 . hence , this correction causes a shift and an asymmetry between high- and low - density regions , with no change in the amplitude at @xmath51 . in particular , when we use a threshold rescaled by a volume fraction of the smoothed field , @xmath47 , the genus of the matter density field per unit volume
expanded to first order in mass variance @xmath20 can be written as a sum of a gaussian term @xmath52 plus a non - gaussian term @xmath53 , namely : @xmath54 the gaussian part is expressed by @xmath55 while the non - gaussian term is given by ( young - rae kim et al . , in preparation ) : @xmath56\sigma_{0}. \nonumber\end{aligned}\ ] ] in the previous equations , @xmath57 are hermite polynomials , and in particular @xmath58 , @xmath59 , @xmath60 , @xmath61 , @xmath62 , and @xmath63 .
also , the various @xmath64 , @xmath65 , are skewness parameters obtained by integrating the bispectrum @xmath66 over @xmath67 and @xmath68 ( see eq .
61 - 64 in matsubara 2003 ) .
in addition , the bispectrum can be given in terms of the nonlinear contributions from nonlinear gravitational evolution , and also primordial non - gaussianity an aspect that we do not consider here ( but see appendix b in hikage et al .
note that the non - gaussian part @xmath53 only appears with terms of the form @xmath69 . assuming galaxy biasing local and deterministic in the weakly nonlinear regime , the skewness parameters of the galaxy bispectrum , @xmath70 , is given by @xmath71 where @xmath72 and @xmath73 are bias parameters . the non - gaussian term @xmath53 in equation [ eq : matsubara ] only appears as combinations of the form @xmath74 when the biased variance at first order is then given by @xmath75 ; therefore
, the non - gaussian correction of the galaxy density field is exactly the same as the one of the unbiased mass density field in equation [ eq : matsubara ] hence independent of the bias parameter .
this can be considered as an advantage of the volume fraction threshold , as opposed to the direct density threshold ( see matsubara 2003 for more details ) .
the measured genus curve can be compared with predictions of the simplest inflationary models , which assume gaussian random phase initial conditions ( and so with eq .
[ eq_genus_standard ] ) .
however , even if the initial conditions were perfectly gaussian , small deviations from gaussianity are expected because of systematic effects ( for example shot noise or redshift - space distortions ) , and because of the physics connected with galaxy formation , nonlinear gravitational evolution , and primordial non - gaussianity ( if any ) .
therefore , it is important to quantify even small departures from gaussianity from the observed genus .
this is done by parametrizing the genus curve with several derived quantities . in what follows ,
we consider measurements as a function of the volume fraction threshold @xmath47 , and introduce four genus - related statistics .
the first quantity is simply the best - fit genus amplitude @xmath76 , measured by a least - squares fit of the theoretical random phase curve to the data considering only the range @xmath77 . in principle
, its value is given by equation [ eq_genus_amplitude ] , but the measured one is always lower because of nonlinear clustering and biasing due to coalescence of structures ( park & gott 1991 ; vogeley et al .
1994 ; canavezes et al . 1998 ; gott et al .
2008 ) .
the second quantity is the shift parameter @xmath78 , defined as @xmath79 where @xmath80 and @xmath81 are the observed and best - fit gaussian genus curves both given by equation [ eq_genus_standard ] , but in the former case with the observed amplitude @xmath82 and in the latter with the best - fit one , @xmath76 , as explained in park et al .
the parameter @xmath78 controls the horizontal shifts of the central part of the genus curve . for a density field dominated by voids , @xmath78 is positive and we say that the density field has a `` bubble - like '' topology . for a cluster - dominated field
, @xmath78 is negative and we say that the field has a `` meatball - like '' topology .
we then further introduce two additional quantities , @xmath83 and @xmath84 , which measure the abundances of clusters ( c ) and voids ( v ) , respectively , relative to the expectations for a gaussian random field .
they are defined by the following relation @xmath85 where the integration intervals are @xmath86 for @xmath87 , and @xmath88 for @xmath84 ( park , gott , & choi 2001 ; park , kim , & gott 2005a ) .
these intervals are centered near the minima of the gaussian genus curve ( i.e. @xmath89 ) , far away from the thresholds where the genus is often affected by the shift phenomenon .
the previous intervals also exclude extreme thresholds , where for low density regions @xmath47 is very sensitive to the exact density value .
these parameters are defined so that the condition @xmath90 ( @xmath91 ) implies that more independent clusters ( voids ) are observed , with respect to those predicted by a gaussian field at a fixed volume fraction . on the opposite
, @xmath92 ( @xmath93 ) means that fewer independent clusters ( voids ) are seen .
the effects of gravitational evolution , galaxy biasing , and cosmology dependence on the statistics defined by @xmath78 , @xmath84 , and @xmath83 as a function of the smoothing scale have been addressed in detail by park , kim , & gott ( 2005a ) ; we refer to their study for more details .
in this section we briefly describe our lrg sample obtained from the sdss dr7 , along with the methodology applied to the observational dataset . the genus computed from the lrg sample and its related statistics
will be presented later on , in section [ sec : genus_top ] .
the sdss is a successful ground - based survey , designed to explore the large - scale distribution of galaxies and quasars by using a dedicated 2.5 m telescope at apache point observatory ( see gunn et al .
2006 for technical details ) .
the photometric survey has imaged roughly @xmath94 steradians of the northern galactic cap in five photometric bandpasses denoted by @xmath95 , and @xmath96 , and centered at 3551 , 4686 , 6165 , 7481 and 8931 angstroms , respectively .
the imaging camera used is equipped with 54 ccds ( fukugita et al .
1996 ; gunn et al .
the limiting photometric magnitudes are 22.0 , 22.2 , 22.2 , 21.3 and 20.5 in the previous five bandpasses , at a signal - to - noise ratio of @xmath97 .
the median width of the psf is @xmath98 and the @xmath21 photometric uncertainties are at the @xmath99 level ( abazajian et al .
2004 ) . after image processing ( lupton et al . 2001
; stoughton et al .
2002 ; pier et al . 2003 ) and calibration ( hogg et al . 2001
; smith et al .
2002 ) , targets are selected for spectroscopic follow - up observations . the spectra are obtained with two dual fiber - fed ccd spectrographs at a spectral resolution @xmath100 , and @xmath21 uncertainty in redshift of @xmath101 km@xmath102 .
because of mechanical constructions , two fibers can not be placed closer than @xmath103 on the same tile . the incompleteness percentage of the spectroscopic survey reaches about @xmath104 .
the sdss spectroscopy yields three major samples : the main galaxy sample ( strauss et al .
2002 ) , the lrg sample ( eisenstein et al . 2001 ) , and the quasar sample ( richards et al . 2002 ) .
in particular , the lrg sample considered here is part of the final data release of the sdss - ii , indicated as dr7 , which yields @xmath105 galaxy spectra over the legacy spectroscopic coverage of @xmath106 deg@xmath107 . in this work
, we made a volume - limited sample including @xmath108 lrgs in the redshift range from 0.16 to 0.36 and rest - frame @xmath10-band absolute magnitudes of @xmath11 , passively evolved to @xmath109 ( see zehavi et al .
2005 ; eisenstein et al . 2005 ) , by using the `` dr7-full '' sample of kazin et al .
k - corrections have been applied to all the galaxies in the sample , assuming a fiducial @xmath4cdm model with @xmath110 and @xmath111 , not @xmath112 which was applied to the sample by kazin et al .
( 2010 ) . to maximize the volume - to - surface ratio
, we trim the sample as in choi et al .
( 2010 ) see their figure 1 for more details .
both the southern galactic cap region and the hubble deep field region are dropped .
these cuts leave a total of @xmath113 lrgs over about 2.33 @xmath114 in the survey region with an angular selection function greater than 0.6 . figure [ fig : isosurface ] shows three - dimensional isodensity contours of the smoothed galaxy number density fields obtained from the sdss lrg sample at @xmath115 and 0 , of which the corresponding volume fractions are 2.3% , 6.7% , and 50% .
a gaussian smoothing is applied with @xmath116 . as expected ,
the asymmetry between high- and low - density regions of the observed genus curve shown in figure [ fig : main_result ] is also clearly seen in this visual comparison .
the low - density regions ( left upper panels ) tend to be more connected and filamentary than the high - density regions ( right upper panels ) , where structures appear to be more isolated and rounder .
as the volume fraction increases , the structures increase in size . for an arbitrary large - scale galaxy survey ,
the sampling of galaxies as a function of redshift is usually not uniform .
moreover , typically the survey is designed so that not only the mean galaxy number density is not constant , but also the sampling in absolute magnitude is non - uniform .
an example is shown in figure [ fig : mz ] , for the semi - volume - limited lrg sample .
the solid lines in the panel identify the volume - limited sample considered in this study .
the high non - uniformity of the sample , as a function of redshift and absolute magnitude , is clearly visible . in this situation
, it would be incorrect to give a single - value weight to galaxies in each redshift bin , based only on the radial selection function ; in fact , this simple scheme would over - weight the galaxies fully sampled , and under - weight those under - sampled .
galaxies with different luminosity are known to cluster differently ( park et al .
1994 ; park et al .
2005b ; zehavi et al .
2005 ; guo et al .
2013 ) , and therefore they should get different weights if the sampling varies with luminosity to avoid the clustering mismatch . the problem becomes more serious when galaxy luminosity or mass are used as weights , to obtain the galaxy luminosity density or the mass density field , respectively .
in particular , when the sampling in luminosity or mass varies with redshift , the resulting luminosity or mass density will have different mean values across different redshift bins , even if the galaxy number density is matched .
therefore , one should also consider the radial density gradient . for the construction of our galaxy mass and number density fields from the observed lrg sample , we devise a new weighting scheme ( called ` luminosity function matching ' ) which properly accounts for the sampling rate variations depending on the location of the galaxy both in redshift and absolute magnitude space , variations that are caused by the lrg target selection procedure . in
what follows , we consider the case when there is no evolution of the luminosity function ( lf ) with redshift , and briefly summarize our procedure ( see also figs . [
fig : lf ] , [ fig : ndenhist ] and [ fig : mag2mass ] below ) .
a. select a reference redshift bin , and compute the reference lf in this bin indicated with @xmath117 .
the lf determined in this ( arbitrary ) redshift interval will be used to match the lf in other redshift bins , as shown in figure [ fig : lf ] . for our study
, we choose the interval @xmath118 as the reference @xmath96-bin ; @xmath117 computed in this interval is indicated with the thick solid line in figure [ fig : lf ] .
b. select bins in the two - dimensional plane defined by redshift versus absolute magnitude , and compute the lf for each pixel of the two - dimensional array @xmath119 above an absolute magnitude cut , @xmath120 . c. for each pixel of the two - dimensional array @xmath119 , calculate the proper weight @xmath121 as @xmath122 from figure [ fig : lf ]
, one can easily infer , just by looking in the magnitude range fainter than @xmath123 , that this weighting scheme will clearly weight galaxies at moderate redshifts more than those at low redshifts .
d. construct the galaxy number density field , weighting each galaxy by @xmath121 , which is linearly interpolated from the two - dimensional array @xmath119 computed as described in the previous steps . the number density field obtained in this way
will be uniform both in redshift and luminosity space ( see the thick blue histogram in fig . [
fig : ndenhist ] ) , as opposed to the one constructed by weighting each galaxy with the radial selection function alone ( dotted line in the same figure ) .
in particular , the density field is calculated on a mesh with cubic pixels from a discrete particle distribution using the cloud - in - cell ( cic ) mass assignment scheme .
e. alternatively , construct the mass - weighted halo density field from the observed galaxy sample .
the galaxy mass , @xmath124 should be the halo mass @xmath125 corresponding to the @xmath10-band galaxy luminosity @xmath126 , i.e. @xmath127 ( see point iii . in section [ sec : mock ] for more details ) . using the lrg cumulative lf measured at the reference redshift bin , and the halo cumulative mass function derived from a full cubic data snapshot of the hr3 at @xmath128 ( which is compatible with the reference redshift )
, we apply the halo - galaxy one - to - one correspondence model ( hgc ) of kim , park , & choi ( 2008 ) and convert galaxy luminosities into halo masses , and vice - versa .
+ figure [ fig : mag2mass ] shows the relation between galaxy luminosity and halo mass , used to determine @xmath124 .
the halo mass corresponding to the absolute magnitude cut , @xmath120 , is @xmath129 . to compute the galaxy mass density field ,
each galaxy is weighted by @xmath130 .
the mass density field derived with this procedure is equivalent to the one obtained from uniformly selected lrgs .
in this section we briefly describe the horizon run 3 @xmath0-body simulation , and the procedure to construct the sdss dr7 mock lrg samples from the simulation output .
we will then compare numerical results and measurements from data in section [ sec : genus_top ] .
the mock surveys will also be used to quantify several nonlinearities due to systematics which affect the genus curve : correcting for these effects allows one to accurately recover the topology of the underlying matter , as we will present in section [ sec : genus_sys ] .
the horizon run 3 ( hr3 ; kim et al .
2011 ) is one of the largest @xmath0-body simulations to date , made using @xmath131 billion particles , spanning a volume of @xmath132gpc)@xmath133 which is over 8800 times the volume of the millennium run ( springel et al . 2005 ) .
the particle mass is down to @xmath134 , allowing to resolve galaxy - size halos with mean particle separation of @xmath135mpc .
the simulation is based on the @xmath4cdm cosmology , with parameters fixed by the wmap 5-year data ( komatsu et al .
the linear power spectrum used is obtained with the camb source ( http://camb.info/sources ) , which provides a better measurement of the bao scale .
the simulation starts at @xmath136 , and is evolved till the present epoch with 600 global time steps .
the code used for the run is an improved version of the grid - of - oct - trees - particle - mesh code ( gotpm ) , originally devised by dubinski et al .
( 2004 ) ; a new procedure has been implemented , in order to describe more accurately the particle positions using single precision . in the hr3
, halos are first identified via a standard friend - of - friend ( fof ) procedure .
then subhalos are found out of fof halos with a subhalo finding technique developed by kim & park ( 2006 ) and kim , park , & choi ( 2008 ) .
this method allows one to identify physically self - bound ( psb ) dark matter subhalos not tidally disrupted by larger structures at the desired epoch .
in particular , lrgs are identified as the most massive dark matter subhalos . to make the comparison with observational data
, we saved the particle positions and velocities along the past light cone for 27 separated observers , and found subhalos in the past light cone surface from @xmath137 to @xmath138 . from each simulated light cone
, we made 3 mock samples using exactly the same survey mask and angular selection function of the sdss sample .
in addition , we applied the same smoothing length as for the observational case . in total , we are able to obtain 81 non - overlapping mock samples , thanks to the enormous volume of the hr3 .
to this end , we note that the ability to simulate big volumes is essential ( particularly for the lrg distribution ) , since larger volumes allow one to model more accurately the true power at large scales and the corresponding power spectrum .
a large box size will guarantee small statistical errors in power spectrum estimates , so that the acoustic peak scale can be measured with an accuracy better than @xmath139 , and the genus curve characterized with unprecedented statistical significance .
a crucial step in our analysis is the construction of realistic mock lrg samples .
this requires the ability to mimic all the observational biases , such as survey boundary , radial and angular selection function , redshift space distortions and so forth . to build the various simulated catalogs ,
27 observers were placed in the hr3 box , each covering the redshift range @xmath140 without overlaps : this means that the survey volumes are totally independent .
in addition , the lrg mocks are made so that they span exactly the same range in absolute magnitude as the observational sample ; hence , the number of galaxies in each mock is nearly equal to the observed one ( at the percentage level accuracy ) .
moreover , the simulated galaxies should be observed in redshift space along the past light cone with the same radial and angular selection function of the observational sample , and also with the same selection function in absolute magnitude space as in the real observation .
figure [ fig : isosurface_mock ] shows a three - dimensional example of the simulated lrg number density field obtained from the hr3 , smoothed with a gaussian filter at @xmath141 scale . the plot is the equivalent of figure [ fig : isosurface ] , but now for the lrg mock samples constructed from the hr3 simulation .
again , the left panels display three representative density contours enclosing low - density regions , which occupy respectively @xmath142 ( @xmath143 ; top ) , @xmath144 ( @xmath145 ; intermediate ) , and @xmath146 ( @xmath147 ; bottom ) of the sample volume . the same thresholds , but now with positive signs and so for high - density regions ( i.e. @xmath148 , top ; @xmath149 , intermediate ; @xmath147 , bottom ) , are shown in the right panels . since we apply identical techniques both to the sdss lrg sample and to the lrg mock surveys
, we expect the results of the analysis to be identical across datasets within statistical variations if the simulations are correctly modeling the distribution of galaxies . in what follows ,
we describe in more detail how to build the sdss dr7 mock lrg samples from the hr3 simulation output .
results from our procedure confirm that we are correctly modeling the lrg distribution ( see figs .
[ fig : lfmock ] and [ fig : ndenhist_mock ] ) .
the major steps of the construction process are summarized next .
a. locate 27 observers in the hr3 simulation box , and save all dark halos along the past light cone of the observer during the simulation , in the redshift range @xmath150 . from each light cone data ,
make 3 mock surveys using exactly the same survey mask and angular selection function as for the sdss volume - limited sample .
b. apply a proper correction to make the halo mass function uniform in redshift .
in fact , in the hr3 simulation the minimum mass limit of subhalos that can have lrgs with constant observed number density varies as a function of redshift ( see fig . 6 in kim
et al . 2011 ) .
this leads to the following relation ; @xmath151 .
the correction one needs to apply to the halo mass at an arbitrary redshift @xmath96 is then given by the ratio @xmath152 , where @xmath153 is the minimum halo mass at a median redshift of @xmath154 in the reference redshift bin ( recall the procedure described in section [ sec : mden ] , and the chosen reference redshift interval ) .
c. populate dark matter halos with galaxies using a suitable correspondence scheme .
in essence , to connect galaxies with halos one needs to make an assumption on the relation between galaxy luminosity and halo mass .
a widely - used approach is the subhalo abundance matching , where more luminous galaxies are assigned to more massive haloes ( kravtsov et al .
2004 ; tasitsiomi et al .
2004 ; vale & ostriker 2006 ; conroy & wechsler 2009 ; guo et al . 2010 ; behroozi et al . 2010 ; kim , park & choi 2008 ) .
this scheme assumes that halos with mass above a certain threshold and with a given mean number density correspond to galaxies with luminosity or mass above a certain threshold and having the same mean halo number density .
for our mocks , we apply the halo - galaxy one - to - one monotonic correspondence model ( hgc ) of kim , park , & choi ( 2008 ) which extends the subhalo abundance matching procedure : there is one and only one galaxy in each subhalo , and a more massive subhalo hosts a more luminous galaxy .
the mapping @xmath127 is shown in figure [ fig : mag2mass ] .
this correspondence scheme allows us to assign a luminosity to each lrg mock galaxy , and to compute galaxy masses ( see also the end of sec . [
sec : mden ] ) .
d. account for the effects of the color - dependent luminosity cut imposed by the sdss lrg volume - limited sample selection criteria , both in redshift and luminosity space , which reduces the sampling density ( see again sec . [
sec : mden ] ) . in order to do so , we discard mock galaxies with a rejection probability given by @xmath155 , where @xmath156 is derived from the observed sample ( see eq .
[ eq : w ] ) .
our procedure successfully reproduces the dependence of the lrg sampling rate on luminosity and redshift as in the sdss lrg sample .
this is shown in figures [ fig : lfmock ] and [ fig : ndenhist_mock ] , the corresponding counterparts of figures [ fig : lf ] and [ fig : ndenhist ] , respectively , obtained from simulated samples .
in particular , figure [ fig : lfmock ] displays the luminosity function computed at different redshift bins from one of our lrg mocks . to facilitate the comparison with the actual sdss data , open circles in the figure are luminosity function measurements derived from the sdss lrg sample in the reference redshift bin @xmath157 as in figure [ fig : lf ] , see section [ sec : mden ] .
clearly , the simulated results agree very well with observational measurements in terms of matching the luminosity function . to this end , figure [ fig : ndenhist_mock ] shows the comoving number density of sdss mock lrgs averaged over all the 81 mock samples , as a function of redshift ( black dashed line ) .
solid line shows the radial distribution obtained by weighting each galaxy with @xmath121 .
colored thin lines show results of six arbitrary mock surveys . filled and open circles in the figure are analogous measurements derived from the sdss lrg sample as in figure [ fig : ndenhist ] .
the weighting scheme used is explained in section [ sec : mden ] . even in this case , the plot confirms the correctness of our modeling procedure : our mocks have the same sampling rate in redshift as the observed sdss lrg sample .
in this section we present results for the genus measured from the sdss dr7 lrg sample , and from our lrg mocks obtained from the hr3 lcdm simulation . in both cases , we compute the genus curves using the mass weighted density field and the number density field although later on we will only use the number density . by contrasting observational results against mock measurements which assume gaussian initial conditions , we detect significant non - gaussian deviations of the observed genus curve from theoretical expectations .
we then further quantify these discrepancies by introducing a new statistical test .
a large part of the non - gaussian deviations is caused by systematics , and we address their impact on the genus curve in section [ sec : genus_sys ] . in sections [ sec : mden ] and [
sec : mock ] we constructed the mass weighted density field and the number density field in order to compute the genus .
this is because jee et al .
( 2012 ) found that the halo mass density has a much tighter ( and simpler ) relation with the underlying matter density than the halo number density .
a similar conclusion was reached by park , kim , & park ( 2010 ) , in relation to the gravitational shear field . to this end
, figure [ fig : mgplot ] shows the genus curves derived from the mass density and number density fields , as a function of the volume fraction , @xmath47 .
no corrections for systematics are yet applied .
we smooth our density field ( either the mass weighted or the number density one ) with a gaussian filter of radius @xmath158 at a pixel size of @xmath159 . to minimize any nonlinearity introduced by the choice of the pixel dimension
, we use the smallest possible pixel size we can afford . in the figure ,
red thick solid lines are used for the number density field , and black thin lines are for the mass weighted density field . in particular
, the top panel shows our measurements averaged from 81 mock lrg samples , where we also plot the genus curve of the dark matter distribution in real space using the full simulation cube with the dotted green line .
the difference between halo and matter density field curves are mostly due to halo biasing and discrete sampling of the halo density field ( i.e. shot noise ) . the genus amplitude measured from the halo mass weighted density field is lower than the one obtained from the halo number density field .
this is consistent with the result of seljak , hamaus , & desjacques ( 2009 ) ; namely , weighting halo galaxies by halo mass can significantly suppress shot noise .
the bottom panel shows similar measurements from the sdss lrg sample . to date , these are the best measurements of the genus curve from the sdss survey catalog . in table [
tab : mgplot_genus ] of appendix [ sec : genus_curves ] , the genus values are given as a function of the volume - fraction threshold level .
in particular , the genus amplitude obtained from the number density field is equal to @xmath160 , with a @xmath161 error including all systematic effects such as finite pixel size , survey boundary , radial and angular selection functions and sparse sampling ( see also fig . [
fig : main_result ] ) .
the errorbars in the figure are the @xmath162 deviations computed from 81 independent mock samples .
although the shape of the genus curve does not depend on the weighting scheme as much as its amplitude , results from simulations reveal that the mass weighted density field tends to show more ` meatball'-shifted topology and more isolated clusters . in table
[ tab : mgplot ] we report the details of the genus - related statistics ( see sec . [ sec : grs ] ) , for the genus measurements displayed in figure [ fig : mgplot ] , while in appendix [ sec : genus_curves ] we list the complete genus values relative to these measurements , as a function of the volume - fraction threshold @xmath47 ( see tab .
[ tab : mgplot_genus ] ) .
we also provide similar measurements as a function of @xmath29
. the latter table may be useful for readers who wish to directly use galaxy clustering topology for cosmological applications
. lllll sample & @xmath76 & @xmath78 & @xmath84 & @xmath83 + dm & @xmath163 & @xmath164 & @xmath165 & @xmath166 + + number & @xmath167 & @xmath168 & @xmath169 & @xmath170 + mass & @xmath171 & @xmath172 & @xmath173 & @xmath174 + + number & @xmath175 & @xmath176 & @xmath177 & @xmath178 + mass & @xmath179 & @xmath180 & @xmath181 & @xmath182 + * notes . * ` number ' and ` mass ' stand for number density field and mass weighted density field , respectively .
@xmath76 is the amplitude of the best - fit gaussian genus curve , @xmath183 is the shift parameter , and @xmath83 and @xmath84 are cluster and void abundance parameters , respectively .
all these values are not bias - corrected .
uncertainty limits are estimated for 81 mock samples .
deviations of the genus curves from the gaussian expectation are quantified by @xmath78 , @xmath84 and @xmath83 .
[ tab : mgplot ] from figure [ fig : mgplot ] and from the results for the genus - related statistics , it is evident that in the mock measurements low - density regions are definitely less affected by the weighting scheme . instead
, the weighting scheme does not introduce any change in the genus amplitude derived from the observational sample , and the mass weighted density field still produces more isolated clusters .
this finding clearly points towards the existence of a significant amount of scatter in the relation between galaxy observables and their underlying halos ; hence , one needs to gain a better understanding of the connection between galaxies and their dark matter halos , and of the galaxy formation process in general . in this paper , hereafter we will adopt the use of number density instead of the mass weighted density , for simplicity .
overall , the hgc galaxy assignment scheme of kim , park , & choi ( 2008 ) is able to match well the observed amplitudes and shapes of the corresponding genus curves .
in fact , from table [ tab : mgplot ] , one can notice that the values obtained for @xmath184 , @xmath84 and @xmath83 from the simulated lrg samples agree well within the quoted uncertainties with those measured from the observational sample .
however , we detect some significant discrepancies between mocks and observations for @xmath84 in the lower density regions beyond the integration intervals quoted in section [ sec : grs ] ( i.e. @xmath185 ) see indeed the difference between the red and grey lines in figure [ fig : main_result ] .
we also detect some discrepancies for the observed and predicted genus amplitudes .
to estimate quantitatively the statistical significance of these discrepancies , we use the following method .
first , we calculate the differences between the genus curve of each individual mock sample and the curve obtained by averaging all the 81 mock samples ; we do this at three different intervals , i.e. @xmath186 , @xmath187 , and @xmath188 .
we then plot the integrals of these differences as histograms in figure [ fig : gstats_nden ] . finally , we place in the same plot our measurements obtained from the sdss sample at the corresponding threshold intervals , indicated with triangular symbols . here
we measured genus curves from the number density fields of the samples .
as one can infer from the figure ( with a significance level of 90% ) , departures from gaussianity are seen near the mean density regions , and in low - density regions ( i.e. in the interval @xmath189 ) .
the radical difference between the genus curves of the observational and simulated data in low - density regions shows that topology is highly sensitive to the connectivity of voids . in the next part
, we will address the key role played by systematics on the genus curve which will explain some of the discrepancies , and show how to accurately correct for their effects to recover the topology of the underlying matter . in a forthcoming paper
, we provide an interpretation of the remaining deviations ( i.e. after correcting for known systematics ) in the context of primordial non - gaussianity .
in this section we briefly discuss the known systematics which affect the genus curve .
we then test and quantify their impact on the genus using the genus - related statistics presented in section [ sec : grs ] , with the help of mock lrg samples ( section [ sec : mock ] ) , and show how to correct for their effects . by applying those corrections , we obtain the most accurate constraint on the genus amplitude to date , which significantly improves on our previous measurements . in particular , figures [ fig : gplot ] and [ fig : gplotdiff ] are among the most important results of our paper . as anticipated in section [ sec : grs ] , even if the initial conditions were perfectly gaussian , small deviations from gaussianity are expected because of systematics . since systematics directly impact the shape and amplitude of the genus curve , it is imperative to be able to quantify and correct for their effects .
this can now be done quite accurately , with the help of realistic mock catalogs such as those constructed from the hr3 ( section [ sec : mock ] ) .
broadly speaking , systematics that cause non - gaussian deviations in the genus curve can be classified into three main classes : those due to the observational or analysis strategy , those due to statistics , and those of cosmological origin .
finite pixel size effects , survey boundary mask , radial and angular selection function , past light cone gradient , and initial conditions of the simulations belong to the first class .
shot noise or sparse sampling and cosmic variance are of statistical origin , while galaxy biasing , nonlinear gravitational evolution , and redshift - space distortions ( rsd ) are related to cosmology . sometimes , but
this depends on the chosen terminology , the last class is not considered as a systematic effect . here
, we broadly term all these three classes as systematic biases on the genus curve .
clearly , several of the previously mentioned effects are connected , and so one needs to remove them simultaneously . in the absence of other known systematics , an eventual residual of non - gaussianity ( after applying the corrections mentioned above ) has to be ascribed to a primordial origin . in what follows ,
we discuss in particular the nonlinear gravitational evolution , and the effects of galaxy bias and past light cone on the genus .
more details on the full modeling of systematics in topology measurements will be presented in young - rae kim et al .
( in preparation ) . in this work
we perform similar corrections as those applied by choi et al .
( 2010 ) , who studied the effects of systematics on the genus computed from the nearby main galaxy sample of the sdss dr7 .
the overall goal is to remove the nonlinear systematics in the observed sample step - by - step , as well as to estimate the genus curve of the underlying matter density field using a set of mock samples .
we first consider the effect of nonlinear gravitational evolution on the genus curve , measured from our mock lrg samples along with survey mask and initial condition effects .
for this purpose , we compute the genus curves of 81 dark matter density fields , both at the initial ( @xmath190 ) and final ( @xmath191 ) redshifts ; these quantities , measured in real space , are indicated as @xmath192 and @xmath193 , respectively , where the index @xmath194 refers to the particular light - cone mock survey considered , @xmath195 stands for matter , and @xmath196 for real space . each density field
is selected within the sdss survey mask , at a particular region in the simulation so that the evolved density field has a one - to - one correspondence with the initial density field .
.genus - related statistics of the samples used in figures [ fig : sysbias_dm ] and [ fig : sysbias_halo ] . [ cols="<,<,>,>,>",options="header " , ] [ tab : gplot_matter ] * notes . * @xmath80 is the genus value of the sdss lrg sample , and @xmath197 is the one averaged over all the 81 light cone lrg mock samples , @xmath198 .
@xmath199 and @xmath200 are the genus values of the underlying dark matter distribution , derived from @xmath80 and @xmath197 , respectively , after correcting for the various systematics . @xmath201 and @xmath202 are the real space genus values of the dark matter distribution at the initial epoch of the simulation , which include only the contribution of primordial non - gaussianity ( both for the observation and simulation ) .
@xmath203 is the genus value for matter distribution in real space at @xmath137 , using the full cubic data .
figure [ fig : gplot_dth ] shows the sdss and mock genus curves , plotted versus the direct density threshold @xmath29 .
these curves deviate from the gaussian predictions more than those obtained using the volume - fraction @xmath47 shown in figure [ fig : main_result ] .
table [ tab : mgplot_genus ] from the previous appendix [ sec : genus_curves ] lists the genus values as a function of @xmath29 . by inspecting those values ,
we conclude that the direct density threshold parametrization is more sensitive to the skewness in the density probability distribution .
however , given the vanishing genus and its large dispersion at threshold levels below @xmath204 ( see figure [ fig : nuerr ] ) , it is not appropriate to use this genus curve to inspect the non - gaussian deviation of the observational sample from perturbation theory , at the smoothing scale adopted .
abazajian , k. , et al .
2004 , , 128 , 502 abazajian , k. n. , et al .
2009 , , 182 , 543 adelman - mccarthy , j. k. , agueros , m. a. , allam , s. s. , et al .
2006 , , 162 , 38 adelman - mccarthy j. k. , et al .
2007 , , 172 , 634 adler , r. j. 1981 , the geometry of random fields , ed .
adler , r. j. ( chichester : wiley ) bardeen , j. m. , bond , j. r. , kaiser , n. , & szalay , a. s. 1986 , , 304 , 15 bassett , b. a. , parkinson , d. , & nichol , r. c. 2005 , , 626 , 1l behroozi , p. s. , conroy , c. , & wechsler , r. h. 2010 , , 717 , 379 blake , c. , et al .
2011 , , 418 , 1707 blanton , m. r. , et al .
2005 , , 129 , 2562 canavezes , a. , et al . 1998 , , 297 , 777 choi , y .- y .
2010 , , 190 , 181 choi , y .- y . , han , d .- h .
, & kim , s. s. 2010 , j. korean astron .
43 , 191 conroy , c. & wechsler , r. h. 2009 , , 696 , 620 doroshkevich , g. 1970 , astrophysika , 6 , 320 dubinski , j. , kim , j. , park , c. , & humble , r. 2004 , new astron , 9 , 111 eisenstein , d. j. , et al .
2001 , , 122 , 2267 eisenstein , d. j. , blanton , m. , zehavi , i. , bahcall , n. , brinkmann , j. , loveday , j. , meiksin , a. , & schneider , d. 2005 , , 619 , 178 fukugita , m. , ichikawa , t. , gunn , j. e. , doi , m. , shimasaku , k. , & schneider , d. p. 1996
, , 111 , 1748 gangui , a. , lucchin , f. , matarrese , s. , & mollerach , s. 1994 , , 430 , 447 gott , j. r. , merlot , a. l. , & dickinson , m. , 1986 , , 306 , 341 gott , j. r. , weinberg , d. n. , & merlot a. l. , 1987 , , 319 , 1 gott , j. r. , et al .
1989 , , 340 , 625 gott , j. r. , et al .
2008 , , 675 , 16 gott , j. r. , et al . , 2009 , , 695 , l45 gunn , j. e. , et al
. 1998 , , 116 , 3040 gunn , j. e. , et al .
2006 , , 131 , 2332 guo , h. , zehavi , i. , zheng , z. , et al .
2013 , , 767 , 122 guo , q. , white , s. , li , c. , & boylan - kolchin , m. 2010 , , 404 , 1111 guth , a. h. 1981 , , 23 , 347 hamilton , a. j. s. , gott , j. r. , & weinberg , d. 1986 , , 309 , 1 hwang , j .- c .
2012 , j. korean astron .
45 , 65 hikage , c.,taruya , a. , & suto , y. 2001 , 556 , 641 hikage , c. , et al .
2002 , , 54 , 707 hikage , c.,taruya , a. , & suto , y. 2003 , 55 , 335 hikage , c. , et al .
2003 , , 55 , 911 hikage c. , komatsu e. , & matsubara t. 2006 , , 653 , 11 hoyle , f. , vogeley , m. s. , & gott , j. r. 2002 , , 570 , 44 hogg , d. w. , finkbeiner , d. p. , schlegel , d. j. , & gunn , j. e. 2001 , , 122 , 2129 james , j. b. , colless , m. , lewis , g. f. , & peacock , j. a. 2009 , , 394 , 454 james , j. b. , lewis , g. f. , & colless , m. 2007 , , 375 , 128 jee , i. , park , c. , kim , j. , choi , y .- y .
, & kim , s. s. 2012 , , 753 , 11 jeong , e. & smoot , g. f. 2007 , arxiv:0710.2371 kazin , e. a. , et al .
2010 , , 710 , 1444 kim , j. , & park , c. 2006 , , 639 , 600 kim , j. , park , c. , & choi , y .- y .
2008 , , 683 , 123 kim , j. , park , c. , rossi , g. , lee , s. m. , & gott , j. r. 2011 , j. korean astronom .
217 komatsu , e. , & spergel , d. n. 2001 , , 63 , 063002 komatsu , e. , smith , k. m. , dunkley , j. , et al .
2011 , , 192 , 18 komatsu , e. , dunkley , j. , nolta , m. r. , et al .
2009 , , 180 , 330 kravtsov , a. v. , berlind , a. a. , wechsler , r. h. , et al .
2004 , , 609 , 35 linde , a. d. 1982 , physics letters b 108 , 389 lupton , r. h. , gunn , j. e. , ivezic , z. , knapp , g. r. , kent , s. , & yasuda , n. 2001 , in asp conf .
238 , astronomical data analysis software and systems x , ed .
f. r. harnden , jr .
, f. a. primini , & h. e. payne ( san francisco : asp ) , 269 matsubara , t. 1994 , , 434 , l43 matsubara , t. 2003 , , 594 , 1 moore , b. , et al .
1992 , , 256 , 477 melott , a. l. , weinberg , d. h. , & gott , j. r. 1988 , , 328 , 50 park , h. , kim , j. , & park , c. 2010 , , 714 , 207 park , c. , pranav , p. , chingangbam , p. , et al .
2013 , journal of korean astronomical society , 46 , 125 park , c. , & gott , j. r. 1991 , , 378 , 457 park , c. , gott , j. r. , & choi , y. j. 2001 , , 553 , 33 park , c. , gott , j. r. , & da costa , l. n. 1992 , , 392 , l51 park , c. , gott , j. r. , melott , a. l. , & karachentsev , i. d. 1992 , , 387 , 1 park , c. , vogeley , m. s. , geller , m. j. , & huchra , j. p. 1994
, , 431 , 569 park , c. , kim , j. , & gott , j. r. 2005a , , 633 , 1 park , c. , et al .
2005b , , 633 , 11 park , c. , & kim y .- r .
2010 , , 715 , l185 park , c. , choi , y .- y . , kim , j. , gott , j. r. , kim , s. s. , & kim , k .- s .
2012 , , 759 , 7l percival , w. j. , et al . 2007 , , 657 , 645 pier , j. r. , munn , j. a. , hindsley , r. b. , hennessy , g. s. , kent , s. m. , lupton , r. h. , & ivezic , z. 2003 , , 125 , 1559 protogeros , z. a. m. , & weinberg , d. h. 1997 , , 489 , 457 rhoads , j. e. , gott , j. r. , & postman , m. 1994 , , 421 , 1 richards , g. t. , et al .
2002 , , 123 , 2945 ross , a. j. , percival , w. j. , snchez , a. g. , et al . 2012 , , 424 , 564 salopek , d. s. , & bond , j. r. 1990 , , 42 , 3936 seljak , u. , hamaus , n. , & desjacques , v. 2009 , , 103 , 091303 slosar , a. , hirata , c. , seljak , u. , ho , s. , & padmanabhan , n. 2008 , j. cosmol .
, jcap08(2008)031 smidt , j. , amblard , a. , byrnes , c. t. , et al .
2010 , , 81 , 123007 smith , j. a. , et al .
2002 , , 123 , 2121 spergel , d. n. , et al .
2003 , , 148 , 175 springel v. et al .
2005 , , 435 , 629 stoughton , c. , et al . 2002 , , 123 , 485 strauss , m. a. , et al . 2002 , , 124 , 1810 tasitsiomi , a. , kravtsov , a. v. , wechsler , r. h. , & primack , j. p. 2004
, , 614 , 533 tegmark , m. , et al .
2006 , phys . rev .
d 74 , 123507 vale , a. , & ostriker , j. p. 2004
, , 353 , 189 verde , l. , wang , l. , heavens , a. f. , & kamionkowski , m. 2000 , , 313 , 141 vogeley , m. s. , park , c. , geller , m. j. , huchra , j. p. , & gott , j. r. 1994 , , 420 , 525 weinberg , d. h. , gott , j. r. , & melott , a. l. 1987 , , 321 , 2 yadav , a. p. s. , & wandelt , b. d. 2008 , , 100 , 181301 york , d. g. , et al .
2000 , , 120 , 1579 zehavi , i. , et al .
2005 , , 621 , 22 | we present measurements of the genus topology of luminous red galaxies ( lrgs ) from the sloan digital sky survey ( sdss ) data release 7 catalog , with unprecedented statistical significance . to estimate the uncertainties in the measured genus , we construct 81 mock sdss lrg surveys along the past light cone from the horizon run 3 , one of the largest @xmath0-body simulations to date that evolved @xmath1 particles in a @xmath2mpc size box .
after carefully modeling and removing all known systematic effects due to finite pixel size , survey boundary , radial and angular selection functions , shot noise and galaxy biasing , we find the observed genus amplitude to reach 272 at 22 @xmath3mpc smoothing scale with an uncertainty of 4.2% ; the estimated error fully incorporates cosmic variance .
this is the most accurate constraint of the genus amplitude to date , which significantly improves on our previous results .
in particular , the shape of the genus curve agrees very well with the mean topology of the sdss lrg mock surveys in the @xmath4cdm universe .
however , comparison with simulations also shows small deviations of the observed genus curve from the theoretical expectation for gaussian initial conditions .
while these discrepancies are mainly driven by known systematic effects such as those of shot noise and redshift - space distortions , they do contain important cosmological information on the physical effects connected with galaxy formation , gravitational evolution and primordial non - gaussianity .
we address here the key role played by systematics on the genus curve , and show how to accurately correct for their effects to recover the topology of the underlying matter . in a forthcoming paper , we provide an interpretation of those deviations in the context of the local model of non - gaussianity . |
high mass x - ray binaries host a compact object ( a neutron star or a black hole ) and a massive star ( liu , van paradijs & van den heuvel 2000 , 2005 , 2006 ) .
the major subclass of high mass x - ray binaries consists of a be star and a compact object and they are referred to as be x - ray binaries ( be xrbs , e.g. , hayasaki & okazaki 2005 ; 2006 ) .
the be stars are massive , generally main sequence , stars of spectral types a0-o8 with balmer emission lines ( zorec & briot 1997 , negueruela 1998 ) .
the be xrbs are found with rather wide ( orbital periods in the range of @xmath5 days ) and frequently eccentric orbit and a compact object accretes from the wind of a be star ( even massive be stars are within their roche lobes for these wide orbits ) . at present , @xmath0 be xrbs are known in the galaxy , and in @xmath1 the compact object was confirmed to be a neutron star ( ns ) by the presence of the x - ray pulsations ( see table [ galactic ] ) . in the remaining cases , whenever we have information concerning the nature of the compact component ( such as an x - ray spectrum ) , it also indicates a ns . although one can not exclude that a few of these systems contain white dwarfs or black holes , it is fair to state that majority of them contain nss as compact components .
other classes of xrbs are ( with one exception ) less numerous . we know
@xmath6 x - ray bursters ( which all host nss ) and @xmath7 x - ray pulsars not associated with a be type companion ( @xmath8 of these nss are associated with a supergiant type companion and @xmath9 with a low mass companion ) .
in addition , we know @xmath10 black hole candidate systems ( among them 21 confirmed bh systems ; e.g. , orosz 2003 ; casares 2007 ; ziolkowski 2008 )
. however , not a single black hole binary containing a be type component has been found so far .
this disparity , @xmath1 be xrbs with nss versus not a single one with a bh , seems indeed striking .
the x - ray emission from be xrbs ( with a few exceptions ) is of a distinctly transient nature with rather short active phases separated by much longer quiescent intervals ( a flaring behavior ) .
there are two types of flares , which are classified as type i outbursts ( smaller and regularly repeating ) and type ii outbursts ( larger and irregular ; negueruela & okazaki 2001 , negueruela et al .
type i bursts are observed in systems with highly eccentric orbits .
they occur close to periastron passages of a ns .
they are repeating at intervals @xmath11 .
type ii bursts may occur at any orbital phase .
they are correlated with the disruption of the excretion disc around be star ( as observed in h@xmath12 line ) .
they repeat on time scale of the dynamical evolution of the excretion disc ( @xmath13 few to few tens of years ) .
this recurrence time scale is generally much longer than the orbital period ( negueruela et al .
2001 ) .
be xrbs systems are known to contain two discs : excretion disc around be star and accretion disc around neutron star .
both discs are temporary : excretion disc disperses and refills on time scales @xmath13 few to few tens of years ( dynamical evolution of the disc , formerly known as the `` activity of a be star '' ( negueruela et al .
2001 ) , while the accretion disc disperses and refills on time scales @xmath13 weeks to months ( which is related to the orbital motion on an eccentric orbit and , on some occasions , also to the major instabilities of the excretion disc ) .
the accretion disc might be absent over a longer period of time ( @xmath13 years ) , if the other disc is very weak or absent .
the x - ray emission of be xrbs binaries is controlled by the centrifugal gate mechanism , which , in turn , is operated both by the periastron passages ( type i bursts ) and by the dynamical evolution of the excretion disc ( both types of bursts ) .
this mechanism explains the transient nature of the x - ray emission ( see ziolkowski 2002 and references therein ) . one should add that the excretion discs are not a mystery any more . in recent years
, the outflowing viscous excretion discs were used to describe the circumstellar matter around be stars known earlier as `` an envelope of a be star '' ( okazaki 1997 ; porter 1999 ; negueruela & okazaki 2001 ) . the modeling with the help of the viscous excretion discs appeared to be by far more successful in describing the circumstellar matter , than earlier descriptions in terms of `` equatorial winds '' , `` expanding envelopes '' or `` ejected shells '' . in particular , the viscous disc models were able to explain the very low outflow velocities ( the observed upper limits are , at most , a few km s@xmath14 and , also , to explain the ( so called ) v / r variability , observed in be stars .
the viscous excretion discs are very similar to the , well known , viscous accretion discs , except for the changed sign of the rate of the mass flow .
some aspects of the modeling ( the supply of the matter with the sufficient angular momentum to the inner edge of the disc , the interaction of the stellar radiation with the matter of the disc ) are not fully solved yet , but the general picture is quite convincing . the viscosity in the excretion discs is usually assumed ( similarly as for accretion discs ) to be in the form of @xmath12-viscosity .
the discs are almost keplerian ( rotationally supported ) which explains the very low values of the radial component of velocity .
nearly keplerian discs ( both inflowing and outflowing ) were , since a long time , known to undergo a global one - armed oscillation instability ( kato 1983 ) .
this instability ( progressing density waves ) provide a very successful explanation of v / r variability , observed both in isolated be stars and in members of be / x - ray systems .
this phenomenon manifests itself in the form of quasi - cyclical changes of the ratio of the strengths of the v(iolet ) peak to the r(ed ) peak in the double profile emission lines .
this variability ( best seen for the h@xmath15 line ) includes phases , when only one peak is visible .
the time scales of the quasi - cycles range from months to years or decades .
the theoretical line profiles calculated for the discs with an asymmetric matter distribution ( due to progressing density waves ) were found to be in good agreement with the observed profiles ( okazaki 1996 ; hummel & hanuschik 1997 ) .
also the theoretical time scales calculated for the one - armed oscillation instability agreed with the observed time scales of v / r variability ( negueruela et al .
the one - armed instability leads ultimately to the disruption of the disc and ejection of the matter from its outer rim .
this phenomenon is believed to be responsible for type ii bursts .
therefore , these time scales describe also the recurrence of type ii bursts .
the total number of be xrbs in the galaxy is difficult to estimate .
the typical duty cycle ( the relative length of the interval of the high x - ray emission , during which the system might be detected ) is rather small : @xmath16 to @xmath17 ( with the exception of a few persistent sources ) .
the coverage of the sky by x - ray surveys is very far from complete .
many systems lie in the obscured regions of the galaxy and they might be detected only in hard x - rays . with the exception of integral ( hard x - ray capability which has led to the discovery of many new be xrbs ; bird et al . 2007 ) , these systems are undetectable by most x - ray observatories . since most of
be xrbs are dormant at any given moment , we can expect to discover many new such systems in near future .
it might be estimated that the total number of be xrbs in the galaxy is perhaps an order or two orders of magnitude larger than the number of the presently known systems ( e.g. , rappaport & van den heuvel 1982 ; van den heuvel & rappaport 1987 ) . in a series of conference proceeding papers ( sadowski et al .
2008a , 2008b , 2008c ) we have reported a potential solution to the problem of the missing black hole be x - ray binaries .
this solution was incorrect and we want to stress to not use these papers .
the errors were made in population synthesis result analysis . in this study , we have performed the calculations anew and we have double checked our analysis .
additionally , all the calculations were performed with the revised code ( see 2.1 ) .
the new results along with the new and qualitatively different solution to the missing problem are presented in this study . in this work
we study the origins of the apparent disparity of number of known be xrbs with nss ( @xmath1 ) as compared to no known be xrbs with bhs in galaxy .
this disparity has been noted in the literature for some time .
the first stellar population synthesis calculations intended to estimate the number of black hole be x - ray binaries were carried out by raguzova & lipunov ( 1999 ) .
they _ assumed _ that a ratio of be x - ray binaries with a ns to the ones with a bh is @xmath18 .
this was based on @xmath19 galactic ns be x - ray binaries known at that time and the assumption that grs 1915 + 105 is a bh be x - ray binary .
we know now that this is not the case and that grs 1915 + 105 is a low mass x - ray binary with the donor star being a low mass ( k - m ) giant ( e.g. , greiner et al . 2001 ) .
they demonstrated that by adjusting some evolutionary parameters ( e.g. , natal kicks , the initial star mass that divides a ns from a bh formation ) their model can reproduce the ratio of @xmath18 .
however , some of their adjustments seem to be rather extreme in the light of the current understanding of binary evolution . for example
, they needed to assume that stars form black holes _ only _ if their mass is higher than @xmath20 and that _ each _ main sequence star above @xmath21 becomes a be star after a mass accretion episode . another population synthesis study connected to the be phenomenon
was presented by zhang et al .
( 2004 ) . in this study
it was noted that , according to the stellar population synthesis calculations by podsiadlowski et al .
( 2003 ) , bh binaries are predominantly formed with relatively short orbital periods ( @xmath22days ) . if this is the case , then , according to zhang et al .
( 2004 ) , the excretion disc truncation mechanism ( artymowicz & lubow 1994 ) might be so efficient that the accretion rate is very low and the system remains dormant ( and therefore invisible ) almost all the time .
one should note , however , that podsiadlowski et al .
( 2003 ) considered essentially bh systems with roche lobe filling secondaries , definitely not the case for be xrbs ( wind - fed accretion ; e.g. negueruela 1998 ; ziolkowski 2002 ) .
we want to stress that in this paper we consider only galactic be xrbs .
however , it should be added that there exists a large population of be xrbs in the magellanic clouds ( e.g. , liu et al .
it is even larger than that of the galaxy : @xmath23 known systems with @xmath24 containing a confirmed pulsar ; ziolkowski , unpublished compilation .
moreover , this population is growing very fast . taking into account that the total stellar content of the magellanic clouds is much smaller than the galactic one
, the study of the origin of this population might be very interesting .
due to very different properties of stellar populations ( e.g. , lower metallicity , different star formation history ) we defer the study of be xrbs in the magellanic clouds to another paper .
binary population synthesis is used to calculate the population of massive stars ( spectral types b / o ) that are in binaries with a compact object ; either a neutron star ( ns ) or a black hole ( bh ) .
the population synthesis code employed in this work , startrack , was initially developed for the study of double compact object mergers in the context of gamma - ray burst ( grb ) progenitors ( belczynski , bulik & rudak 2002a ) and gravitational - wave inspiral sources ( belczynski , kalogera & bulik 2002b ) .
single stellar evolution is based mostly on the modified set of stellar models of hurley , pols & tout ( 2000 ) while binary evolution procedures were independently developed for this code . in recent years startrack
has undergone major updates and revisions in the physical treatment of various binary evolution phases , and especially mass transfer phases . the new version has already been tested and calibrated against observations and detailed binary mass transfer calculations ( belczynski et al .
2008 ) , and has been used in various applications .
the physics updates that are most important for compact object formation and evolution include : ( i ) a full numerical approach to orbital evolution due to tidal interactions , calibrated using high mass x - ray binaries and open cluster observations ; ( ii ) a detailed treatment of mass transfer episodes fully calibrated against detailed calculations with a stellar evolution code ; ( iii ) updated stellar winds for massive stars ( see belczynski et al .
2009a ) ; and ( iv ) the latest determination of the natal kick velocity distribution for neutron stars ( hobbs et al .
2005 ) . for he
star evolution , which is of crucial importance for the formation of double neutron star binaries ( e.g. ivanova et al .
2003 ; dewi & pols 2003 ) , we have applied a treatment matching closely the results of detailed evolutionary calculations .
if the he star expands and roche lobe overflow ( rlof ) begins , the systems are examined for the potential development of a dynamical instability , in which case they are evolved through a common envelope ( ce ) phase , otherwise a highly non - conservative mass transfer ensues .
we treat ce events using the energy formalism ( webbink 1984 ) , where the binding energy of the envelope is determined from the set of he star models calculated with the detailed evolutionary code by ivanova et al .
the progenitor evolution and roche lobe overflow episodes are now followed in much greater detail .
in particular the code was calibrated for various cases of mass transfer in massive binaries and then tested against some published detailed evolutionary tracks ( e.g. wellstein , langer & brown 2001 ) . one specific evolutionary process that plays an important role for the formation of binaries with black holes is the ce phase .
this phase is instrumental in decreasing orbital separation and bringing initially distant binary components close to each other .
this in turn allows components of many systems to begin an interaction and manifest their presence through a number of observational phenomena .
the inspiral process in common envelope and the subsequent decrease of an orbit is not well understood .
it is believed that only donor stars ( initiating the process ) with the well developed core - envelope structure , like red giants , can successfully survive the inspiral . on the other hand stars with no clear core - envelope boundary , like main sequence stars , can not survive this phase , leading to a ce merger and formation of one , rapidly rotating and peculiar star .
it was pointed out that stars crossing the hertzsprung gap ( hg ) do not have a clear entropy jump at the core - envelope transition ( ivanova & taam 2004 ) .
therefore , it can be expected that if such a star overflows its roche lobe and initiates a ce phase the inspiral will eventually lead to a merger ( e.g. taam & sandquist 2000 ) .
this possibility was tested in evolutionary calculations of merger rates of double compact objects .
it turns out that if such a possibility is accounted for , the merger rates of bh - bh binaries decrease drastically ( by factor of @xmath25 ) .
ns - ns merger rates drop only moderately ( belczynski et al . 2007 ) .
since the details of the ce phase are not fully understood , in this study we will also test this possibility in our predictions for be x - ray binaries . in model
a , we will allow for survival even if a donor star is on the hg ( i.e. , standard energy balance is calculated to check on the outcome of ce ) , while in model b , any ce phase that involves an hg donor is _ assumed _ to lead to a merger aborting subsequent binary evolution and potential formation of a be x - ray binary .
another important factor that determines the formation efficiencies of binaries with neutron stars and black holes is the natal kick compact objects receive at birth . for models
a and b we adopt natal kicks from the radio pulsar birth velocity distribution derived by hobbs et al .
( 2005 ; a maxwellian with @xmath26 km s@xmath14 ) for neutron stars that are formed in regular core - collapse supernovae , while for neutron stars formed in electron capture supernovae we adopt no natal kicks ( e.g. dessart et al .
natal kicks for black holes are decreased proportionally to the amount of fall back expected during core collapse / supernova explosion ( e.g. fryer 1999 ; fryer & kalogera 2001 ) .
however , we calculate one extra model in which natal kicks are reduced .
in particular , in model c , neutron star kicks are drawn from a maxwellian distribution with @xmath27 km s@xmath14 , while black hole kicks are derived from the same distribution but then decreased due to the fall back . for the electron capture neutron star formation
, no natal kicks are applied .
there are some indications that the natal kicks neutron stars receive are smaller for stars in binaries as compared with single stars ( e.g. podsiadlowski et al .
2004 ) . here
we have decreased the value of kicks as measured by hobbs et al .
( 2005 ) for single pulsars by half and applied them to neutron stars in binaries .
this specific choice was motivated by the recent calculation that demonstrated that in order to reproduce the ratio of recycled pulsars in double neutron star binaries ( 9 known ) to single recycled pulsars ( 4 ) the kicks operating in binary stars are required to be rather low ( @xmath28 km s@xmath14 ; belczynski et al .
2009b ) .
delineation in the formation of neutron stars and black holes may play an important role in the expected number of be x - ray binaries with different accretors .
in particular , we still do not know what a star in the initial mass range of @xmath29 forms : a neutron star or a black hole ?
in this study we have employed a physical model for the formation of compact object , and rather than using a star initial mass ( as most population synthesis codes do ) , we have used the star properties to delineate the formation of a neutron star from the formation of a black hole . in short ,
we follow the evolution of a given star , note the final mass of its feni core and then we use the results of hydro core - collapse simulations to estimate the mass of the compact object a given star forms .
once we have the mass of a compact object we use the maximum neutron star mass ( assumed to be @xmath30 ) to tell apart neutron stars from black holes .
the full description of this scheme is given in belczynski et al .
( 2008 ; and see the references within ) . as it happens our scheme ( for single stars ) results in the black hole formation for @xmath31 ( see belczynski et al .
if we wanted to ( artificially ) adopt a higher limit ( e.g. , @xmath32 ) , we would expect more be x - ray binaries with neutron stars than predicted in the following sections .
in other words , our conclusions are rather conservative on the estimate of the ratio of be x - ray binaries with neutron stars to the ones with black holes . if anything such a ratio should be higher and our conclusions that follow stronger .
we evolve a galactic population of massive binaries using startrack .
we adopt solar metallicity ( @xmath33 ) , and a steep initial mass function ( imf ) for massive stars with a power - law exponent of @xmath34 ( kroupa , tout & gilmore 1993 ; kroupa & weidner 2003 ) .
roche lobe overflow is treated in a non - conservative way ( with @xmath35 mass loss from a given binary ; e.g. meurs & van den heuvel 1989 ) while the ce phase is treated via energy balance with fully efficient transfer of orbital energy into dispersal of an envelope ( e.g. @xmath36 ) .
the results are calibrated in such a way that the galactic star formation rate is at the level of @xmath37 and is constant through the last 10 gyr ( e.g. oshaughnessy et al .
2006 ) . at the present galactic disk age ( @xmath38 gyr )
we perform a time slice and extract be x - ray binaries using classification criteria defined in the following section . following the earlier discussion ( 1 ) of the observed properties of galactic population of
be x - ray binaries , we adopt a definition ( that is extended to potential systems that may host a bh ) of a be x - ray binary in our population synthesis calculations .
we call any system a be x - ray binary if : _
( i ) _ it hosts either a ns or a bh accretor ; _ ( ii ) _ donor is a main sequence star ( burning h in its core ) ; _ ( iii ) _ donor mass is high @xmath39 ( o / b star ) ; _ ( iv ) _ accretion proceeds _ only _ via stellar wind ( no rlof ) ; _ ( v ) _
orbital period is in the range @xmath40 day ; and _ ( vi ) _ only a fraction @xmath41 of the above systems are designated as hosting a be star and not a regular o / b star .
the observations indicate that the fraction of be stars among all b stars is @xmath42 to @xmath43 ( e.g. , slettebak 1988 ; ziolkowski 2002 ; mcswain & gies 2005 ) .
note that in the above definition we do not require a non - zero eccentricity as for the galactic bexrbs as many as @xmath44 systems have small eccentricity ( @xmath45 ) .
we have chosen to define be x - ray binary in a phenomenological way , and based on the observational properties we have adopted the values for our limiting factors ( e.g. , @xmath46 ) .
note also , that we do not have put any constrains on the spin of a be star .
it translates into an assumption ( as be stars are known to have high rotational velocities ; e.g. , slettebak 1949 , 1966 ; slettebak , collins & truax 1992 ) that either some stars are born with initially high spins or that the mass accretion during rlof can effectively spin up a massive star . the more physical approach
( e.g. , b star becomes be star if spun up via rlof accretion ) is also potentially possible within the framework of our population synthesis model .
however , the details of such a model would be highly uncertain ( e.g. , the high spin of be stars may be connected to their initial rapid rotation and not to rlof spin up ; it is not certain if any spin up is expected during ce phase ; and how effectively angular momentum can be transferred during rlof ) .
we note that the adopted value of @xmath46 affects only the absolute number of be x - ray binaries predicted in our calculations .
however , the physical properties or the relative ratio of be x - ray binaries hosting a ns to the ones hosting a bh remains the same and therefore our results and conclusions ( see below ) are independent of the value we adopt for @xmath46 .
in table [ channels ] , the main results of our calculations are presented .
the main formation channels for be x - ray binaries are given separately for systems with nss ( marked @xmath47 where @xmath48 is a number that indicates a given channel ) and bhs ( @xmath49 ) .
we also list the predicted intrinsic galactic number of be x - ray binaries ( @xmath50 ) , along with the expected ratio of systems with nss to systems with bhs ( @xmath51 ) .
binaries with ns form mostly along two channels . one ( @xmath52 ; that forms @xmath53 of be x - ray binaries with nss ) involves a ce phase when the primary is an evolved star ( e.g. burning he in its core , or burning h and he in a shell ) and after losing the envelope , the combination of the post - ce orbital separation ( rather large ) and maximum radius of exposed primary core ( not too large ) is such that there are no more interactions until the primary s exposed core explodes and forms a ns . the other channel ( @xmath54 ;
@xmath53 ) involves also the initial ce phase , but it usually starts on a smaller orbit , so the post - ce separation is such that it allows the exposed core of primary ( naked he star ) to overflow its roche lobe and start another interaction ( non conservative mass transfer ) before it explodes and forms a ns . there is also one channel ( @xmath55 ; @xmath17 ) that involves two binary components of comparable mass , and once rlof begins it does not develop into a ce phase but proceeds via regular and not too violent way ( on a thermal and/or nuclear timescale of the donor ) .
after the rlof episode the primary explodes and forms a ns . at the time of ns formation
the companion is a still unevolved ( main sequence ) star with high mass ( @xmath56 ) .
there is not much difference from model a to b in the formation of be x - ray binaries with nss .
the formation of be x - ray binaries with bhs proceeds along two channels .
the first one ( @xmath57 ) is very similar to @xmath52 and involves a ce phase followed by the explosion of a primary and the formation of a bh .
the main difference comes from the fact that in this case the primary is more massive ( than a progenitor of a ns in @xmath52 ) and it starts a ce phase earlier in its evolution .
the ce phase is encountered very often when a primary is crossing the hg , as it is the first evolutionary phase at which the stars undergo a very significant radial expansion . since we impose non survival in such a case for our model b , we note the significant decrease of formation of be x - ray binaries in this channel from model a ( @xmath58 ) to model b ( @xmath17 )
. the second formation channel ( @xmath59 ) of be x - ray binaries with bhs is almost the same as the channel @xmath55 for binaries with ns .
since this channel involves only one non - conservative rlof , the change from model a ( @xmath44 ) to model b ( @xmath60 ) is only relative .
this change simply reflects the decrease of number of binaries in channel @xmath57 and the actual number of binaries in the channel @xmath59 does not change from model a to b. we also present the predicted numbers of be x - ray binaries calibrated for the galactic disk star formation rate . we have selected ( randomly ) only @xmath61 ( @xmath41 ) of the massive binaries hosting b / o star with a compact object companion as this seems to be the fraction of be stars among regular b stars and we have additionally imposed some constraints on orbital period and system configuration ( see 2.3 ) .
the number of predicted be x - ray binaries with nss is @xmath62 and does not depend much on the adopted model of ce evolution .
the number of be x - ray binaries with bhs is significantly smaller and is sensitive to the adopted ce model ; we find @xmath63 and @xmath64 systems for model a and model b , respectively .
this is qualitatively the same trend as found by belczynski et al .
( 2007 ) , who demonstrated that ce survival / non - survival affects mostly systems with bhs . in each model
we predict more be xrbs with nss than with bhs .
in general there are more nss than bhs ( @xmath65 ; the ratio that depends on adopted imf and the details of the transition from ns to bh formation ) .
additionally , binaries with initially comparable mass components ( like progenitors of be x - ray binaries with nss ) are more likely to survive the first interaction ( usually ce phase ) than the binaries with components of very different masses ( like for many progenitors of be x - ray binaries with bhs ) .
the number of be xrbs with nss is about the same in both models , but the number of binaries with bhs decreases significantly from model a to b. therefore , the ratio of be xrbs with nss to the ones with bhs changes significantly from model a ( @xmath66 ) to model b ( @xmath67 ) . in figs .
[ perioda ] and [ periodb ] we present the distributions of orbital periods and eccentricities for the synthetic be x - ray binaries for models a and b. due to the small number of binaries predicted in the galaxy , the distributions are constructed from the entire population of potential be x - ray binaries ( i.e. @xmath68 ) .
the actual population of synthetic galactic be x - ray binaries as listed in table [ channels ] ( i.e. @xmath41 ) is constructed by random drawing from the distributions shown in figs . [ perioda ] and [ periodb ] .
orbital periods are contained within @xmath40 days by our adopted definition of be x - ray binaries ( see 2.3 ) .
the distributions within these limits are falling off with increasing periods and this general trend holds for both be x - ray binaries with nss and bhs .
the relatively large numbers of systems with small periods are the outcome of the orbital contraction at ce phase .
this is indirectly , but very clearly , demonstrated on the example of be x - ray binaries with bhs in model b ( fig .
[ periodb ] ) in which the majority systems do not evolve through a ce phase and there are almost no binaries with small periods ( e.g. , with @xmath69 days ) .
the eccentricity distributions are similar for models a and b. for be x - ray binaries with nss there is an overall trend of distributions with rising eccentricity .
this is an effect of natal kicks that are rather high ( @xmath70 km s@xmath14 ) so binaries in general receive high kicks and gain significant eccentricity after the first supernova .
it is noted that we only plot here the tail of entire population of initial binaries that may have became be x - ray binaries , but were disrupted in the first supernova . actually , the disruption at the first supernova for massive binaries is very high ( @xmath71 ) and only a very few binaries survive .
there is also an accumulation of systems with very small eccentricities ( @xmath72 ) and this is due to the population of neutron stars that form through electron capture supernovae for which we have assumed no natal kicks , so the non - zero ( but small ) eccentricities for these systems arise from mass loss only .
the general trends are very similar for distributions of eccentricities for be x - ray binaries with bhs .
however , we note two natural changes in relative strengths of the trends as compared to distributions of binaries with nss . first , the increase of distribution with eccentricity is smaller for binaries with bhs , as these receive , on average , lower natal kicks than nss ( due to the fall back and smaller explosion energies in bh formation ) .
second , the contribution of bh binaries with small eccentricities is higher than for binaries with nss .
the most massive black holes form through direct collapse ( or almost full fall back ) for which we assume no ( or rather small ) natal kicks and since there is no ( or almost no ) mass loss , these systems end up with zero ( or very small ) eccentricities .
in model c we have decreased the natal kicks nss and bhs receive and this model is similar to model b as in this calculation we do not allow for the ce survival if a donor is a hg star .
the formation channels and corresponding efficiencies are about the same as for model b. however , as intuitively expected , the formation of binaries with nss increases ( @xmath73 ) significantly as compared to other models ( see table [ channels ] ) . since most progenitors of be xrbs are disrupted at the formation of a compact object ( via supernova natal kick ) once the kicks are decreased the formation is more efficient .
the same effect is found for be xrbs with bhs ( @xmath74 ) , although at a lesser degree as bh kicks are assumed to be smaller and therefore the disruptions are not as important a factor as for binaries with nss .
this trend significantly increases number of be xrbs with nss and only moderately increases the number of the binaries with bhs .
it therefore has a deep impact on the expected ratio of one population to the other .
in particular , the ratio of be xrbs with ns to ones with bhs is found very high for model c : @xmath75 . in fig .
[ periodc ] we show period and eccentricity distributions for model c.
we have performed a population synthesis study of galactic be x - ray binaries . in particular ,
we have attempted to explain the problem of the missing be x - ray binaries with bhs . the only known be x - ray binaries with a confirmed type of compact object host nss .
specifically , we know 64 be x - ray binaries in the galaxy , but only 42 of these systems are known to host a ns .
none of the observed be x - ray binaries hosts a bh .
* previously , we have reported ( sadowski et al .
2008a , 2008b , 2008c ) the potential solution to the problem of the missing be x - ray binaries .
this solution was incorrect and the mistake was due to some errors in population synthesis data analysis .
the new results ( double - checked and obtained with the updated code ) are presented below . *
we predict that both population of be x - ray binaries should exist in the galaxy : those with nss as well as these with bhs . *
the predicted number of be x - ray binaries with nss is much higher ( factors of @xmath2 ) than those with bhs . *
if we use the preferred evolutionary models ( @xmath76 ; models b and c ) we predict that in the observed sample of be x - ray binaries of 42 systems with nss , one should expect only @xmath4 systems with bhs .
it is quite possible that none are yet observed ( small statistics ) . * due to a very low number of expected binaries with bhs , it is very likely that there is no problem with missing be x - ray binaries with bhs in galaxy .
we would like to thank anonymous referee , duncan lorimer , thomas maccarone and andrzej zdziarski for a number of useful comments on this study . additionally , kb is indebted to karolina tarczynska for her unceasing physical rehabilitation support that made the work over this study possible .
jz acknowledges the support from the polish ministry of science and higher education ( mshe ) project 362/1/n - integral ( 2009 - 2012 ) and kb acknowledges the support from mshe grant n n203 302835 ( 2008 - 2011 ) .
artymowicz , p. , & lubow , s.h .
1994 , , 421 , 651 beklen , e. , & finger , m.h .
2009 , atel , 1956 belczynski , k. , bulik , t. , & rudak , b. 2002a , , 571 , 394 belczynski , k. , kalogera , v. , & bulik , t. 2002b , , 572 , 407 belczynski , taam , r. , kalogera , v. , rasio , f. , & bulik , t. 2007 , apj , 662 , 504 belczynski , k. , kalogera , v. , rasio , f. , taam , r. , zezas , a. , bulik , t. , maccarone , t. , & ivanova , n. 2008 , , 174 , 223 belczynski , k. , bulik , t. , fryer , c. , ruiter , a. , vink , j. , & hurley , j. 2009a , , submitted ( arxiv:0904.2784 ) belczynski , k. , lorimer , d. , ridley , j. , & curran , s. 2009b , , submitted ( arxiv:0907.3486 ) bird , a. j. , malizia , a. , bazzano , a. , barlow , e. j. , bassani , l. , hill , a. b. , b?langer , g. , capitanio , f. , clark , d. j. , dean , a. j. , & 13 coauthors 2007 , , 170 , 175 casares , j. 2007 , in black holes from stars to galaxies across the range of masses . edited by v. karas and g. matt .
proceedings of iau symposium 238 , 3 dessart , l. , burrows , a. , ott , c.d . , livne , e. , yoon , s.y . , & langer , n. 2006 , , 644 , 1063 dewi , j.d.m .
, & pols , o.r . 2003 , , 344 , 629 fryer , c. 1999 , , 522 , 413 fryer , c. , & kalogera , v. 2001 , , 554 , 548 greiner , j. , cuby , j. , mccaughrean , m. , castro - tirado , a. , & mennickent , r. 2001 , , 373 , l37 grundstrom , e. , caballero - nieves , s. , gies , d. , huang , w. , mcswain , m. , rafter , s. , riddle , r. , williams , s. , & wingert , d. 2007 , , 656 , 437 hayasaki , k. , & okazaki , a.t .
2005 , , 360 , l15 hayasaki , k. , & okazaki , a.t .
2006 , , 372 , 1140 van den heuvel , e. , & rappaport , s. 1987 , in proceedings of the 92 iau colloquium `` physics of be stars '' ( editors a.slettebak and t.p.snow ) , cambridge university press , p.291 hobbs , g. , lorimer , d.r . ,
lyne , a.g .
, & kramer , m. 2005 , , 360 , 974 hummel , w. , & hanuschik , r.w .
1997 , a&a , 320 , 852 hurley , j. r. , pols , o. r. , & tout , c. a. 2000 , mnras , 315 , 543 ivanova , n. , belczynski , k. , kalogera , v. , rasio , f. , & taam , r. e. 2003 , , 592 , 475 ivanova , n. , & taam , r. e. 2004 , , 601 , 1058 kato , s. 1983 , pasj , 35 , 249 kroupa , p. , tout , c.a .
, & gilmore , g. 1993 , , 262 , 545 kroupa , p. , & weidner , c. 2003 , , 598
, 1076 liu , q.z . , van paradijs , j. , & van den heuvel , e.p.j .
2000 , , 147 , 25 liu , q.z . , van paradijs , j. , & van den heuvel , e.p.j . 2005 , , 442 , 1135 liu , q.z . , van paradijs , j. & van den heuvel , e.p.j .
2006 , a&a , 455 , 1165 markwardt , c.b . , pereira , d. , ray , p.s . ,
smith , e. , & swank , j.h .
2008 , atel , 1679 mcswain , m.v . , & gies , d.r .
2005 , , 161 , 118 meurs , e.j.a .
, & van den heuvel , e.p.j .
1989 , , 226 , 88 mangano , v. 2009 , atel , 1896 negueruela , i. 1998 , a&a , 338 , 505 negueruela , i. , & okazaki , a.t .
2001 , a&a , 369 , 108 negueruela , i. , okazaki , a.t . ,
fabregat , j. , coe , m.j . ,
munari , u. , & tomov , t. 2001 , a&a , 369 , 117 okazaki , a.t .
1996 , pasj , 48 , 305 okazaki , a.t . in _ the interactions of stars with their environment
toth , l.v . ,
kun , m. , & szabados , l. 1997 , konkoly observatory , budapest , 407 orosz , j.a .
2003 , in a massive star odyssey : from main sequence to supernova , proceedings of iau symposium 212 , p.365 oshaughnessy , r. , kim , c. , kalogera , v. , & belczynski , k. 2006 , , 672 , 479 podsiadlowski , p. , langer , n. , poelarends , a.j.t . , rappaport , s. , heger , a. , & pfahl , e.d .
2004 , , 612 , 1044 podsiadlowski , ph . ,
rappaport , s. , & han , z. 2003 , , 341 , 385 porter , j.m .
1999 , a&a , 348 , 512 raguzova , n. , & lipunov , v. 1999 , a&a , 349 , 505 raguzova , n.v .
2007 , be star newsletter , 38 , 24 reig , p. , & zezas , a. 2009 , atel , 2085 rappaport , s. , & van den heuvel , e. 1982 , in proceedings of the symposium `` be stars '' ( editors m.jaschek and h.g.groth ) , dordrecht , d. reidel publishing co. , p. 327
sadowski , a. , ziolkowski , j. , belczynski , k. , & bulik , t. 2008a , proceedings of the 7th integral workshop ( http://pos.sissa.it/cgi-bin/reader/conf.cgi?confid=67 ) , p.76 sadowski , a. , ziolkowski , j. , belczynski , k. , & bulik , t. 2008b , a population explosion : the nature & evolution of x - ray binaries in diverse environments .
aip conference proceedings , volume 1010 , p.407 sadowski , a. , ziolkowski , j. , belczynski , k. , & bulik , t. 2008c , the art of modeling stars in the 21st century , proceedings of the international astronomical union , iau symposium , volume 252 , p.399 slettebak , a. 1949 , , 110 , 498 slettebak , a. 1966 , , 145 , 126 slettebak , a. 1988 , pasp , 100 , 770 slettebak , a. , collins , g. , & truax , r. 1992 , , 81 , 335 smith , d.m . ,
hazelton , b. , coburn , w. , boggs , s.e .
, fivian , m. , hurford , g.j . ,
hudson , h.s .
, grefenstette , b. , & gilmore , r. 2005 , atel , 557 taam , r. e. , & sandquist , e. l. 2000 , , 38 , 113 tavani , m. , & arons , j. 1997 , , 477 , 439 tomsick , j.a . , chaty , s. , rodriguez , j. , walter , r. , & kaaret , p. 2007
, atel , 1231 torii , k. , sugizaki , m. , kohmura , t. , endo , t. , & nagase , f. 1999 , , 523 , l65 webbink , r. f. 1984 , , 277 , 355 wellstein , s. , langer , n. , & braun , h. 2001 , a&a , 369 , 939 zhang , f. , li , x .- d .
, & wang , z .- r . 2004 , , 603 , 663 ziolkowski , j. 2002 , mem .
ital . , 73 , 1038 ziolkowski , j. 2008 , chinese journal of astronomy and astrophysics supplement , 8 , 273 zorec , j. , & briot , d. 1997 , , 318 , 443 2s 0053 + 604 & 203.59 & & 0.26 & @xmath77 & b0.5 ve & 1,2 + 4u 0115 + 634 & 24.3 & 3.61 & 0.34 & @xmath78 & b0.2 ve & 1,2 + igr j01363 + 6610 & & & & @xmath79 & b1 ve & 1,2 + rx j0146.9 + 6121 & & 1404.2 & & @xmath80 & b1 ve & 1,2 + igr j01583 + 6713 & & & & & be & 1 + 1e 0236.6 + 6100 & 26.496 & & 0.55 & @xmath81 & b0 ve & 1,2,3 + v 0332 + 53 & 34.25 & 4.4 & 0.37 & @xmath82 & o8.5 ve & 1,2 + 4u j0352 + 309 & 250.3 & 837.0 & 0.11 & @xmath83 & o9.5 iiie - b0 ve & 1,2 + rx j0440.9 + 4431 & & 202.5 & & @xmath84 & b0.2 ve & 1,2 + exo 051910 + 3737.7 & & & & @xmath79 & b0 ivpe & 1,2 + 1a j0535 + 262 & 111.0 & 103.4 & 0.47 & @xmath85 & o9.7 iiie &
1,2,4 + 1h 0556 + 286 & & & & & b5ne & 1 + igr j06074 + 2205 & & & & & b0.5 ve & 1,5 + sax j0635.2 + 0533 & 11.2 & 0.0338 & & @xmath86 & b1 iiie - b2 ve & 1,2 + xte j0658 - 073 & & 160.4 & & @xmath87 & o9.7 ve & 1,2 + 3a j0726 - 260 & 34.5 & 103.2 & & @xmath88 & o8 - 9 ve & 1,2 + 1h 0739 - 529 & & & & & b7 iv - ve & 1 + 1h 0749 - 600 & & & & & b8 iiie & 1 + rx j0812.4 - 3114 & 81.3 & 31.8851 & & @xmath89 & b0.2 i ve & 1,2 + gs 0834 - 430 & 105.8 & 12.3 & 0.12 & @xmath90 & b0 - 2
iii - ve & 1,2 + gro j1008 - 57 & 247.5 & 93.5 & 0.66 & @xmath91 & b0e & 1,2 + rx j1037.5 - 5647 & & 862.0 & & @xmath92 & b0 iii - ve & 1,2 + 1a 1118 - 615 & & 407.68 & & @xmath93 & o9.5 ve & 1,2,6 + igr j11435 - 6109 & 52.46 & 161.76 & & & be & 1,7 + 2s 1145 - 619 & 187.5 & 292.4 & @xmath94 0.5 & @xmath95 & b0.2 iiie & 1,2 + 1h 1253 - 761 & & & & & b7 vne & 1 + 1h 1255 - 567 & & & & & b5 ve & 1 + 4u
1258 - 61 & 132.5 & 272.0 & @xmath94 0.5 & @xmath96 & b0.7 ve & 1,2 + 2rxp j130159.6 - 635806 & & 704.0 & & @xmath97 & be ? & 1,2 + sax j1324.4 - 6200 & & 170.84 & & & be ? & 1 + 1wga j1346.5 - 6255 & & & & @xmath98 & b0.5 ve & 1,2 + 2s
1417 - 624 & 42.12 & 17.6 & 0.446 & @xmath99 & b1 ve & 1,2 + sax j1452.8 - 5949 & & 437.4 & & @xmath100 & be ?
& 1,2 + xte j1543 - 568 & 75.56 & 27.12 & @xmath101 0.03 & @xmath102 & be ? & 1,2 + 2s 1553 - 542 & 30.6 & 9.26 & @xmath101 0.03 & @xmath103 & be ? & 1,2 + igr j15539 - 6142 & & & & @xmath104 & b2 - 3 vne & 1,2 + igr j16207 - 5129 & & & & @xmath105 & b8 iiie & 1,2 + swift j1626.6 - 5156 & & 15.37 & & & be & 2 + ax j170006 - 4157 & & 714.5 & & @xmath106 & be ?
& 1,2,8 + rx j1739.4 - 2942 & & & & & be ? & 1 + rx j1744.7 - 2713 & & & & @xmath107 & b0.5 v - iiie & 1,2 + [ galactic ] ax j1749.2 - 2725 & & 220.38 & & @xmath108 & be ? & 1,2 + gro j1750 - 27 & 29.8 & 4.45 & & & be ? & 1 + 1xmm j180816.8 - 191940 & & & & @xmath109 & be ? & 2 + ax j1820.5 - 1434 & & 152.26 & & @xmath110 & o9.5 - b0 ve & 1,2 + xte j1824 - 141 & & 120.0 & & & be ? & 9 + 1xmm j183327.7 - 103523 & & & & @xmath111 & b0.5 ve & 2 + 1xmm j183328.7 - 102409 & & & & @xmath112 & b1 - 1.5 iiie & 2 + gs j1843 + 00 & & 29.5 & & @xmath78 & b0 - 2 iv - ve & 1,2 + 2s 1845 - 024 & 242.18 & 94.8 & 0.88 & @xmath113 & be ?
& 1,2 + xte j1858 + 034 & & 221.0 & & & be ? & 1 + 4u 1901 + 03 & 22.58 & 2.763 & 0.036 & @xmath114 & be ?
& 1,2 + xte j1906 + 09 & 28.0 ? & 89.17 & & & be ? & 1 + 1h 1936 + 541 & & & & & be & 1 + xte j1946 + 274 & 169.2 & 15.8 & 0.33 & @xmath115 & b0 - 1 iv - ve & 1,2 + ks 1947 + 300 & 40.415 & 18.76 & 0.03 & @xmath116 & b0 ve & 1,2 + w63 x-1 & & 36.0 & & & be ? & 1 + exo 2030 + 375 & 46.02 & 41.8 & 0.41 & @xmath117 & b0 ve & 1,2 + rx j2030.5 + 4751 & & & & @xmath118 & b0.5 v - iiie & 1,2 + gro j2058 + 42 & 55.03 & 198.0 & & @xmath119 & o9.5-b0
iv - ve & 1,2 + sax j2103.5 + 4545 & 12.68 & 358.61 & @xmath13 0.4 & @xmath120 & b0 ve & 1,2 + 1h 2138 + 579 & & 66.33 & & @xmath121 & b0 - 2 iv - ve & 1,2,10 + 1h 2202 + 501 & & & & & be & 1 + sax j2239.3 + 6116 & 262.6 & 1247.0 & & @xmath122 & b0 v - b2 iiie & 1,2 + bens:01 & 44.2 ( 41.8 ) [ 45.3 ] % & ce : a@xmath123b , sn : a + bens:02 & 42.3 ( 43.9 ) [ 45.0 ] % & ce : a@xmath123b , nc : a@xmath123b , sn : a + bens:03 & 11.9 ( 13.3 ) [ 8.8 ] % & nc : a@xmath123b , sn : a + bens:04 & 1.6 ( 1.0 ) [ 0.9 ] % & all other + & & + bebh:01 & 79.6 ( 13.2 ) [ 17.2 ] % & ce : a@xmath123b , sn : a + bebh:02 & 19.8 ( 85.5 ) [ 82.8 ] % & nc : a@xmath123b , sn : a + bebh:03 & 0.6 ( 1.3 ) [ 0.0]% & all other + & & + n@xmath124 & 579 ( 517 ) [ 1578 ] & galactic number of ns be xrbs + n@xmath125 & 82 ( 19 ) [ 29 ] & galactic number of bh be xrbs + f@xmath126 & 7 ( 27 ) [ 54 ] & number ratio of ns to bh be xrbs + [ channels ] ) of potential be x - ray binaries , and the actual synthetic galactic population as presented in table [ channels ] was obtained by drawing only @xmath127 of systems from these distributions ( i.e. @xmath41 ) . ] but for model b in which survival through a ce phase with a hg donor is not allowed .
note the significantly smaller number of be - bh x - ray binaries , and moderately smaller number of be - ns x - ray binaries . ]
but for model c in which the natal kicks are decreased by half as compared with models a and b. in this model , the survival through a ce phase with a hg donor is not allowed .
note the significantly larger number of be - ns x - ray binaries , and the moderately larger number of be - bh x - ray binaries . ] | in the galaxy there are @xmath0 be x - ray binaries known to - date . out of those ,
@xmath1 host a neutron star , and for the reminder the nature of a companion is not known .
none , so far , is known to host a black hole .
there seems to be no apparent mechanism that would prevent formation or detection of be stars with black holes .
this disparity is referred to as a missing be black hole x - ray binary problem .
we point out that current evolutionary scenarios that lead to the formation of be x - ray binaries predict that the ratio of these binaries with neutron stars to the ones with black holes is rather high @xmath2 , with the more likely formation models providing the values at the high end .
the ratio is a natural outcome of _ (
i ) _ the stellar initial mass function that produces more neutron stars than black holes and _ ( ii ) _ common envelope evolution ( i.e. a major mechanism involved in the formation of interacting binaries ) that naturally selects progenitors of be x - ray binaries with neutron stars ( binaries with comparable mass components have more likely survival probabilities ) over ones with black holes ( which are much more likely to be common envelope mergers ) . a comparison of this ratio ( i.e. @xmath3 ) with the number of confirmed be
neutron star x - ray binaries ( @xmath1 ) indicates that the expected number of be black hole x - ray binaries is of the order of only @xmath4 .
this is entirely consistent with the observed galactic sample . |
galaxies are not distributed randomly across the sky . at least at low redshifts
, they appear to trace distinct patterns : galaxy clusters are connected to each other by long , filamentary structures of galaxies , interspersed with large voids in which few or no galaxies are seen .
a plausible theoretical motivation has arisen for the formation of such large - scale structures ( lss ) of galaxies , namely that the galaxies are tracing an underlying distribution of dark matter . in their most modern form
, models for the formation of these large scale structures postulate that the evolution of the dark matter density field is inextricably linked to the formation and evolution of the galaxies themselves ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these models generally invoke some variation of the biased hierarchical paradigm , in which overdensities , or ` halos ' , in the dark matter distribution undergo successive mergers over time to build halos of increasing mass , with galaxies forming from the baryonic matter in these halos . from these models , and recent observations ,
it is clear that the relationship between the properties of galaxies , and the properties of the dark matter halos in which they reside , is subtle , and is an area in which observational constraints are particularly valuable in constraining models .
we would like to know , observationally , what sort of galaxy occupies what sort of halo as a function of redshift , and how the properties of galaxies change with both redshift and the masses of their parent halos .
one method that has proven especially useful in providing such observational constraints is measuring clustering amplitudes .
fundamentally , the biased hierarchical paradigm requires that overdensities of dark matter should themselves cluster together on the sky , with the strength of clustering depending on their mass @xcite . in principle
, we can make great strides in understanding the relationship between galaxies and the underlying dark matter distribution by measuring the clustering strength of galaxies selected in a particular way , and relating this to theoretical predictions for halo clustering @xcite .
such observations are , however , not straightforward to perform .
reliable clustering measurements require large , homogeneous samples of sources selected over large enough areas of sky to sample a range of dark matter density regimes . ideally , we would like such observations to be performed in the near- and mid - infrared : the ( restframe ) near - infrard is most sensitive to evolved stars and so can pick up large samples of passively evolving systems , whereas the mid - infrared is sensitive to the dusty , active sources in which the stars and central black holes in ( at least some ) passively evolving systems are thought to form . infrared observatories available up to now , however , have not been capable of mapping large enough areas to the required depths to find sufficient numbers of sources , or in enough bands to even crudely discriminate between different populations .
the launch of the _ spitzer _ space telescope @xcite offers the potential to overcome these problems , due to its ability to map large areas of sky in the infrared to greater depths than any previous observatory , and in multiple bands so that dusty , active systems can be differentiated from passively evolving systems .
spitzer _ wide - area infrared extragalactic ( swire ) survey @xcite is the largest of the _ spitzer _ space telescope s six cycle 1 legacy programmes .
the survey covers a total area of 49 square degrees , split between six fields , in all seven of _
spitzer s _ imaging bands ( 3.6160 ) .
the area and depth of swire combine to produce a survey of significant comoving volume , 0.2@xmath12 gpc@xmath13 over @xmath14 , and spatial scales of @xmath15 mpc at @xmath16 .
a principal goal of swire is to study the clustering behaviour of a variety of extragalactic populations . in @xcite
we presented the first detection of galaxy clustering in the survey , measuring a two - point angular correlation function at 3.6 from our validation data , and in @xcite we presented results on the clustering of ultraluminous infrared galaxies ( ulirgs ) at @xmath17 .
@xcite presented angular correlation functions at 3.68.0 from the 4-square degree _ spitzer _ first look survey . in this paper
we extend the analysis of the 3.6- clustering to larger scales and fainter flux limits ( higher redshifts ) .
we begin with a summary of definitions and formalisms in section [ sec : defs ] , then in section [ sec : sample ] we discuss the sample selection , including star / galaxy separation and the angular selection function . in section [ sec : corrfunc ] we present our measurements of the two - point angular correlation function .
we compare our results with previous measurements in the k - band ( section [ sec : kband ] ) and with the galics semi - analytical simulations ( section [ sec : galics ] ) .
the angular clustering amplitudes are used to estimate the spatial correlation lengths , which are then compared with simple parameterized models of clustering evolution , the galics simulations and results from the literature ( section [ sec : r0 ] ) .
section [ sec : conclude ] draws together some conclusions from our analysis .
we use @xmath18 km s@xmath19 mpc@xmath19 with @xmath20 and @xmath21 .
magnitudes are in the ab system unless otherwise noted .
the spatial two - point correlation function @xmath22 is defined through the joint probablility @xmath23 dv_1 dv_2 \label{eq : dpxi}\ ] ] of finding a galaxy in the volume element @xmath24 and a second galaxy in the volume element @xmath25 separated by a distance @xmath26 at a redshift @xmath27 , where @xmath28 is the mean number density of sources ( e.g. , * ? ? ?
* ) . in comoving coordinates
, the correlation function can be parameterized as @xmath29 where @xmath30 measures the strength of the clustering at @xmath31 , @xmath32 measures the scale - dependence and @xmath33 parameterizes the evolution with redshift ( e.g. , * ? ? ?
* ; * ? ? ?
several special values of @xmath33 have particular interpretations .
( 1 ) @xmath34 is the stable clustering model , where the correlation function is fixed in proper coordinates and clustering grows stronger as the background mass distribution expands with the universe .
( 2 ) @xmath35 is the comoving case , where clustering remains constant in comoving coordinates and simply expands with the universe . ( 3 ) @xmath36 is the linear growth model , which corresponds to the application of linear perturbation theory to a scale - free power spectrum in an einstein - de sitter universe .
we note that these models are qualitative indicators of possible evolution scenarios , rather than realistic clustering models @xcite .
the angular two - point correlation function @xmath37 is a measure of the number of pairs of galaxies with separation @xmath38 compared with that expected for a random distribution .
it is defined through the joint probability @xmath39 d\omega_1 d\omega_2 \label{eq : dptheta}\ ] ] of finding a galaxy in solid angle @xmath40 and a second galaxy in solid angle @xmath41 separated by an angle @xmath38 , where @xmath42 is the mean number density of sources ( per steradian ) in the survey ( e.g. , * ? ? ?
if @xmath37 is zero , the distribution of galaxies is unclustered .
the angular correlation function , @xmath37 , is the projection along the line of sight of the spatial correlation function , @xmath22 , and can be calculated from limber s equation @xcite .
if @xmath22 is parameterized as a power - law , as above , then @xmath37is also a power - law @xmath43 the amplitude , @xmath44 , of the angular correlation function can be expressed ( following , e.g. , * ? ? ?
* ) as @xmath45 ^ 2 } \label{eq : limber}\ ] ] where @xmath46/2 ) \over \gamma(\gamma/2 ) } \label{eq : limberfactor}\ ] ] with @xmath47 being the standard gamma function . here ,
@xmath48 is the redshift distribution , @xmath49 is the angular diameter distance , and @xmath50 is the hubble parameter , @xmath51 ( where we have neglected the radiation energy density ) . the spatial correlation length , @xmath30 , can thus be calculated from the amplitude of the angular correlation function ( eq . [ eq : limber ] ) if the redshift distribution , @xmath48 , of the sources in the survey is known .
the swire - en1 field has an area of @xmath529 deg@xmath53 and is coincident with one of five fields observed as part of the european large - area iso ( infrared space observatory ) survey , elais @xcite .
the nominal field centre is @xmath54 @xmath55 ( j2000 ) .
the field was mapped by _ spitzer _ at 3.6 , 4.5 , 5.8 and 8.0 with the infrared array camera ( irac , * ? ? ?
* ) on 2004 january 1420 , and at 24 , 70 and 160 with the multiband imaging photometer ( mips , * ? ? ?
* ) on 2004 january 2128 and 2004 july 29 .
the data can be retrieved from the _ spitzer _ archive with a program identification ( pid ) of 185 , and the enhanced data products ( image mosaics and catalogues ) are available from the _ spitzer _ science center .
full details of the observations and data processing are given in @xcite ; here we summarize the essential details .
the swire - en1 field was mapped by irac with a large grid of pointings , and at each grid point two 30-second images were taken , each one consisting of multiple dithered exposures .
the entire grid was repeated in two epochs , offset by half an array width .
thus , for any point on the sky there are a minimum of four independent sightings ( images ) , and these sightings occur on widely spaced parts of the detector array in order to minimize instrumental signatures .
the entire survey has a minimum depth of four coverages , equal to 120 seconds of exposure time . in some areas
this can be as high as sixteen coverages , or 480 seconds .
the irac data were reduced and flux - calibrated by the _ spitzer _ science center .
further processing of the individual images removed a number of effects ( mostly due to bright objects , primarily stars ) that remained in the pipeline products @xcite .
the images were then coadded into sixteen large mosaics ( or ` tiles ' ) of approximately @xmath56 square degrees each .
sources were detected and their photometry measured with the sextractor package @xcite .
we used the flux measured in a circular aperture of 38 diameter , unless the source was significantly extended , in which case we used the flux within the kron aperture .
the source catalogue was a superset of the swire data release 2 @xcite , containing fainter objects than published at that time .
we analyzed eleven flux - limited samples selected at 3.6 from the swire - en1 catalogue ( table [ tab : amplitudes ] ) .
the deepest sample contained sources with flux densities @xmath57 @xmath1jy(or @xmath58 mag ) , this limit being defined by the flux density at which the differential completeness is approximately 50 per cent .
at this level , the integral completeness is 82 per cent ( see fig .
[ intcomplete ] and discussion in section [ sec : selection ] ) .
flux intervals of @xmath59 or @xmath60 were used , corresponding to intervals of @xmath61 in the median redshift of the samples ( see section [ sec : zdists ] ) .
.angular and spatial clustering strengths for each of the samples .
@xmath62 are the flux limits , @xmath44 are the amplitudes of the angular correlation functions and @xmath63 are the integral constraints .
@xmath64 are the median redshifts and @xmath30 are the spatial correlation lengths , derived from the galics redshift distributions for the @xmath6515.9 @xmath1jy samples and from the impz redshift distributions for the brighter samples ( see section [ sec : r0 ] ) .
[ cols="^,^,^,^,^ " , ] [ tab : lum ] so far we have interpreted the change in @xmath30 as evolution in the clustering strength , but another possibility is that we are looking at different populations of sources in the different samples .
for example , the fainter samples may be dominated by less massive galaxies which are located in halos of lower mass and so are intrinsically less clustered ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ?
we have investigated this possibility using the 3.6- luminosity ( @xmath66 ) as a proxy for stellar mass . in fig .
[ fig : lumdist ] we plot the luminosity distributions of each sample from the @xcite and @xcite models , showing the contribution from early - type ( elliptical / lenticular ) and late - type ( spiral , irregular , starburst ) populations separately . for 3.6- luminosities below @xmath67 l@xmath68
, the two models predict very similar distributions for the total counts , but at higher luminosities the @xcite models have an extended tail of luminous late - type galaxies that are not present in the @xcite models . at bright flux limits ( @xmath69 @xmath1jy ) , the late - types are about 3040 per cent of the total number of galaxies .
the fraction of late - types then increases with decreasing flux limit , until in the faintest sample the late - types are 50 per cent of the total according to the @xcite model and 75 per cent of the total in the @xcite model . in all cases , the redshift distributions of the early - type galaxies
are weighted towards higher redshifts than those of the late - type galaxies . in fig .
[ fig : r0lum ] we plot the spatial correlation lengths against the average 3.6- luminosities for the two models , and tabulate the luminosities in tab . [
tab : lum ] . for the brightest samples ,
i.e. the lowest median redshifts , the average luminosity is approximately constant , at @xmath70 . at fainter flux limits (
@xmath71 @xmath1jy and below ) , it is seen that there is a trend for @xmath30 to decrease with decreasing average luminosity .
recalling that there is a progressively larger fraction of late - type galaxies as the flux limit decreases , both the decrease in average luminosity ( mass ) and the weaker clustering could be a consequence of the changing galaxy population
i.e. an increase in the number of lower - mass late - type galaxies .
however , with the data we have available we can not unambiguously disentangle evolution in the clustering strength from evolution in the relative populations of early - types and late - types .
, as a function of 3.6 luminosity , @xmath72 , derived from the ( * ? ?
* stars ) and ( * ? ? ?
* circles ) model redshift distributions .
the symbol size is proportional to the median redshift of each sample .
luminosity errors are the standard deviations of the luminosity distributions . ] finally , we have compared the best estimates of the correlation lengths ( table [ tab : amplitudes ] ) with values of @xmath30 from the literature .
this is shown in fig .
[ fig : r0z ] where we plot @xmath30 as a function of redshift for our data , together with a variety of galaxy and agn correlation lengths , and several models , compiled by @xcite .
the swire sources are the most clustered population of galaxies at @xmath73 , with only clusters of galaxies exceeding the correlation lengths of the swire samples .
the solid lines are the halo models of @xcite which allow us to estimate the approximate masses of the halos that host the 3.6- selected sources . at @xmath74
the inferred halo masses are approximately constant with redshift with @xmath75 .
the correlation lengths have been shown above to decrease with increasing redshift beyond @xmath76 and this corresponds to a decrease in halo mass from @xmath77 m@xmath68 to @xmath78 m@xmath68 at @xmath79 .
this is consistent with the idea that we are detecting more late - type galaxies , which have lower masses than the early - types that are prevalent in the brighter samples . , of our samples ( red circles ) as a function of redshift , compared with values from the literature ( * ? ? ? * and references therein ) .
the solid lines are the models of @xcite for halos of fixed mass , with masses as indicated .
also shown are indicative @xmath33 models of clustering evolution : stable and linear models with @xmath80 mpc , corresponding to the average of our low-@xmath27 results ; and a linear model normalized to the dark matter clustering strength , @xmath81 mpc , of @xcite . ]
the swire survey has allowed us to measure the angular clustering of sources selected at 3.6 over larger scales and with greater significance than previously has been possible .
the two - point angular correlation function has been measured in eleven flux - limited samples down to @xmath82 @xmath1jy , corresponding to median redshifts @xmath83 .
these angular results are in good agreement with the galics simulations and with @xmath84-band clustering measurements .
we have used limber s equation to infer the spatial correlation lengths from the angular measurements and four estimates of the redshift distributions of each sample .
we find that the systematic uncertainty in the @xmath48 distribution dominates the statistical errors , and adopt the photometric redshift distribution from impz as the best estimate of @xmath48 , extrapolating to higher redshifts using the galics results .
the comoving correlation length varies slowly around @xmath8 mpc out to a median redshift of @xmath76 and then decreases with increasing redshift .
if this change at higher redshifts is due to evolution in the clustering , then the required evolution is very strong ( with @xmath85 mpc and @xmath86 ) and these swire galaxies would be in environments that will evolve into massive clusters by the present day .
there is some indication from the @xcite and @xcite phenomological models that the change in correlation length is due to a decrease in the average 3.6- luminosity ( or equivalently , stellar mass ) of the fainter samples rather than evolution in the clustering .
this decrease in average luminosity is the result of an increased fraction of late - type galaxies in the faint samples .
comparing these swire data with the halo models of @xcite suggests that the fainter samples are selecting sources in lower mass halos .
these latter two results provide a consistent picture where lower - mass late - type galaxies are preferentially found in lower mass halos and thus have weaker clustering .
the comparison of our data with the galics simulations has shown that these models are a good match to the angular clustering , but their redshift distributions differ markedly from the phenomological models and the photometric redshift distributions .
this highlights the need for a better understanding of the spatial clustering and its evolution .
higher - resolution numerical models and improved semi - analytic models will allow us to directly explore the relation between the clustering of the galaxies and the dark matter , i.e. the bias .
for example , the dark matter millenium simulation @xcite has a lower mass resolution limit than galics , and combined with semi - analytic models of the galaxies it is a better match to the area and depth of swire than the current models .
future work will explore the comparison between these models and the swire data presented here .
the pair - counting code npt was provided by the auton lab at carnegie mellon university s school of computer science ( http://www.autonlab.org ) and we thank andrew moore et al . for making this software public .
this work was funded by pparc research grants to seb oliver .
support for this work , part of the _ spitzer _ space telescope legacy science programme , was provided by nasa through a contract issued by the jet propulsion laboratory , california institute of technology under a contract with nasa .
gonzalez - solares , e. , et al . 2004a , mnras , 352 , 44 gonzalez - solares , e. , et al .
2004b , mnras , 358 , 333 granato , g. l. , lacey , c. g. , silva , l. , bressan , a. , baugh , c. m. , cole , s. , frenk , c. s. 2000 , apj , 542 , 710 gray , a. g. , moore , a. w. , nichol , r. c. , connolly , a. j. , genovese , c. , wasserman , l. 2004 , in f. ochsenbein , m. allen and d. egret ( eds ) , astronomical data analysis software and systems xiii , asp conf series , 314 , 249 surace , j. a. , et al . 2005 , ` the swire data release 2 : image atlases and source catalogs for elais - n1 , elais - n2 , xmm - lss and the lockman hole ' , spitzer science center , california institute of technology ( pasadena , ca ) | we investigate the clustering of galaxies selected in the 3.6 band of the _ spitzer _ wide - area infrared extragalactic ( swire ) legacy survey .
the angular two - point correlation function is calculated for eleven samples with flux limits of @xmath0400 @xmath1jy , over an 8 square degree field .
the angular clustering strength is measured at @xmath2-@xmath3 significance at all flux limits , with amplitudes of @xmath4@xmath5 at one degree , for a power - law model , @xmath6 .
we estimate the redshift distributions of the samples using phenomological models , simulations and photometric redshifts , and so derive the spatial correlation lengths .
we compare our results with the galics ( galaxies in cosmological simulations ) models of galaxy evolution and with parameterized models of clustering evolution .
the galics simulations are consistent with our angular correlation functions , but fail to match the spatial clustering inferred from the phenomological models or the photometric redshifts .
we find that the uncertainties in the redshift distributions of our samples dominate the statistical errors in our estimates of the spatial clustering . at low redshifts ( median @xmath7 )
the comoving correlation length is approximately constant , @xmath8 mpc , and then decreases with increasing redshift to a value of @xmath9 mpc for the faintest sample , for which the median redshift is @xmath10 .
we suggest that this trend can be attributed to a decrease in the average galaxy and halo mass in the fainter flux - limited samples , corresponding to changes in the relative numbers of early- and late - type galaxies .
however , we can not rule out strong evolution of the correlation length over @xmath11 .
[ firstpage ] galaxies : evolution galaxies : statistics infrared : galaxies large - scale structure of universe |
if sneutrinos of the mssm are lighter than the lightest chargino and next - to - lightest neutralino , their decays to charged particles are of higher order and strongly suppressed .
therefore sneutrinos decay completely invisibly into a neutrino and the lightest supersymmetric particle ( lsp , the lightest neutralino in the considered case ) . as a result
, they can not be directly detected in cascade decays at the lhc and a threshold scan at the ilc is precluded for the very same reason .
in such a case , the only possibility to discover sneutrinos is to select a well - reconstructible process where the sneutrino is exchanged or shows up as part of a cascade decay , and in which the precise determination of kinematic distributions gives an access to the sneutrino mass .
this idea has already been exploited in ref .
@xcite , where the lepton - sneutrino decay channel of charginos produced at the ilc has been investigated .
it was argued that background effects are sufficiently under control , so a precise mass determination is possible .
here we report on results of our study for determining sneutrino properties in the sps1a scenario assuming a realistic ilc environment which includes beamstrahlung , initial - state radiation , a complete account of reducible backgrounds from sm and susy processes , and a complete matrix - element calculation of the susy signal encompassing all irreducible background and interference contributions @xcite using the event generator ` whizard ` @xcite .
the sneutrino mass can be determined by investigating the chargino decay modes @xmath0 , @xmath1 , in chargino pair production process at the ilc . in the following
we assume that the first and second generation sneutrinos , @xmath2 and @xmath3 , are mass degenerate .
large backgrounds from same - flavor lepton and slepton production can be avoided by selecting events with leptons of different flavor , i.e. one of the charginos decaying to an electron , the other to a muon .
therefore , we search for ( semi-)exclusive final states : [ eq : mainproc ] e^+e^- ^+_1 ^-_1 _ e^*e^-_^+ e^-^+ |_e _ ^0_1 ^ 0_1 .
in the sps1a scenario @xmath4 gev , @xmath5 gev and the born cross section for chargino pair production is predicted to be 173.6 fb at a cm energy of 500 gev . with a br(@xmath6)=13%
, a sufficient sample of events can be expected in this channel .
the two - body chargino decay leads to a uniform decay lepton energy spectrum with edges determined by the chargino and sneutrino masses , which provides a method for their experimental determination @xcite .
experimentally , the signature for the signal process , eq .
( [ eq : mainproc ] ) , is one electron , one anti - muon and missing energy , as a final signature or use both channels . ]
@xmath7 the same final state , eq .
( [ eq : mainproc ] ) , can originate from other production processes as well as from interference terms , e.g. multi - peripheral diagrams , chargino decays into neutralino and @xmath8 , smuon @xmath9 and selectron @xmath10 pair production , and single - resonant sm di - boson production etc .
they will be called collectively as an ` irreducible background ' , in contrast to the ` reducible ' susy background that includes more neutrinos in the final state .
the latter contribution come from processes that lead to one or two leptonically decaying @xmath11 s .
they mainly originate from the production of mixed neutralino pairs @xmath12 , stau pairs as well as chargino pairs .
the latter undergo the subsequent decays to stau and lsp or tau and sneutrino .
.cross sections for signal and relevant background processes for an ilc @xmath13 gev .
isr and beamstrahlung are always included . the ` presel . '
column includes a @xmath14 cut for the final electron to cut out collinear regions , and for the @xmath15 sm as well as all @xmath16-induced processes a @xmath17 cut for particles vanishing in the beampipe .
the last column shows cross sections after the cuts . in parentheses
are the ` whizard ` integration errors .
[ tab:500gevxsec ] [ cols="^,<,>,>",options="header " , ] as seen in tab .
[ tab:500gevxsec ] , the most severe background ( that exceeds the signal by a factor @xmath18 ) comes from gamma - induced @xmath19 and @xmath11 pairs which produce a soft @xmath20 system with small total transverse momentum @xmath21 and large angular separation @xmath22 .
therefore , a cut on transverse momentum of leptons and a cut on azimuthal separation of @xmath23 and @xmath19 reduces gamma - induced backgrounds by 3 to 5 orders of magnitude .
the other large source of sm background @xmath24 pair production can be reduced by applying an upper cut on the lepton energy , since most leptons from @xmath8 decays have energies larger than 40 gev .
we also require two well - visible leptons in the central part of the detector .
all cuts are summarized in tab .
[ tab : cutsall ] .
these cuts are less effective in suppressing the susy background processes , since leptons originating from decays of heavy susy states , with almost degenerate spectrum like sps1a @xcite , are rather soft and angularly uncorrelated . evt / bin + lepton energy distributions after applying the cuts : sm background ( light grey ) , susy background ( medium grey , red ) signal ( dark grey , blue ) for the electrons ( left panel ) and muons ( right panel ) .
we use @xmath25 ab@xmath26 .
, title="fig : " ] lepton energy distributions after applying the cuts : sm background ( light grey ) , susy background ( medium grey , red ) signal ( dark grey , blue ) for the electrons ( left panel ) and muons ( right panel ) .
we use @xmath25 ab@xmath26 .
, title="fig : " ] -0.5 cm
the most straightforward way to determine the mass of otherwise invisible sneutrinos is the measurement of the total cross section for the process eq .
( [ eq : signature ] ) .
a fit of the collider data to samples of mc data for varying sneutrino mass ( assuming masses of other particles known ) allows one to determine the sneutrino mass with an accuracy of @xmath27 gev .
a less model - dependent possibility is to use the on - shell kinematics of chargino decay , where the information about particle masses is encoded in the positions of lower and upper edges of the lepton energy spectra .
the energy spectra of electrons and muons , including isr and beamstrahlung , after cuts are shown in fig .
[ fig : lep500 ] .
the lepton spectrum from sm processes is flat and can easily be subtracted . on the other hand , leptons from the signal and susy background
exhibit a similar energy dependence .
the lower edge of the lepton spectrum is essentially smeared out , mostly due to @xmath21 and @xmath22 cuts which are necessary to kill the photon - induced sm background . the upper edge , however , remains pronounced .
if the chargino mass is determined from elsewhere ( e.g. the threshold scan ) , the upper edge allows one to determine the sneutrino mass with an accuracy of roughly @xmath28 .
the most sophisticated method of mass determination by fitting the entire ` experimental ' lepton energy distributions to the mc samples generated for varying sneutrino mass allows one to go down with the sneutrino mass error to @xmath29 gev .
the chargino pair production process with the subsequent two - body leptonic decay at the ilc with @xmath13 gev has been investigated including all relevant sm and susy backgrounds as well as isr and beamstrahlung . with the event generator ` whizard ` , the full matrix element for production and decay together with off - shell and interference effects were taken into account .
we have demonstrated that this process can allow us to determine the sneutrino mass with the accuracy well below @xmath28 . in scenarios where sneutrinos are otherwise invisible ,
this is the only method of their detection and mass measurement .
the expected high experimental precision at the ilc , however , calls for more precise theoretical methods , so a full nlo calculation of the signal process is a possible future improvement of the present study @xcite .
work partially supported by the german bmbf grant no .
05ha6vfb , the dfg sfb / tr9 `` computational particle physics '' , the helmholtz - alliance `` physics at the terascale '' , the eu network mrtn - ct-2006 - 035505 tools and precision calculations for physics discoveries at colliders " , the ec programme mtkd - ct-2005 - 029466 `` particle physics and cosmology : the interface '' , as well as by the helmholtz - gemeinschaft grant no .
vh - ng-005 at an early stage of the work .
jk is grateful to the theory division for the hospitality extended to him at cern .
j. kalinowski , w. kilian , j. reuter , t. robens and k. rolbiecki , jhep * 0810 * , 090 ( 2008 ) [ arxiv:0809.3997 [ hep - ph ] ] ; t. robens , j. kalinowski , k. rolbiecki , w. kilian and j. reuter , acta phys . pol .
b * 39 * , 1705 ( 2008 ) [ arxiv:0803.4161 [ hep - ph ] ] .
http://whizard.event-generator.org ; m. moretti , t. ohl , j. reuter , hep - ph/0102195 ; w. kilian , lc - tool-2001 - 039 ; w. kilian , t. ohl , j. reuter , to appear in jhep , arxiv:0708.4233 [ hep - ph ] . j. a. aguilar - saavedra _ et al .
_ , eur . phys .
j. c * 46 * , 43 ( 2006 ) [ arxiv : hep - ph/0511344 ] .
w. kilian , j. reuter and t. robens , eur .
j. c * 48 * , 389 ( 2006 ) ; t. robens , phd thesis , 2006 , arxiv : hep - ph/0610401 ; k. rolbiecki , arxiv:0710.1748 [ hep - ph ] . | using the event generator ` whizard ` we study in a realistic ilc environment the prospects of measuring properties of sneutrinos that decay invisibly into the lightest neutralino and the neutrino . |
about sixty years ago , newton and wigner showed that there is no position operator for a massless particle with spin higher than @xmath0 @xcite .
this fact lead many authors to conclude that it is not possible to define a wave function for a photon , which has zero mass and spin ( or helicity ) 1 .
however , bialynicki - birula @xcite and , independently , sipe @xcite , introduced a function of the position and time coordinates that completely describes the quantum state of a photon . in their opinion , which is shared by us , such function may be referred to as the photon wave function .
the wave equation for this function can be derived from the einstein kinematics for a particle with spin 1 and zero mass in the same way that the dirac equation is obtained for a massive particle with spin @xmath0 @xcite . a strong argument in favor of this photon wave function formulation is that the corresponding wave equation is completely equivalent to the maxwell equations in vacuum . in the present work
we extend the bialynicki - birula sipe formalism to include the interaction of photons with non - absorptive continuous media . in section [ sec : vac ] we briefly discuss the photon wave equation in vacuum and some properties of photon wave functions . in section [ sec : medium ] the photon wave equation is modified including a term associated with the induced current density in the medium , that makes it equivalent to the macroscopic maxwell equations for electromagnetic waves in the presence of matter . in section [ sec : secquant ] we apply the second quantization procedure to the photon wave function formalism in a linear non - dispersive and non - absorptive medium , allowing for description of quantum phenomena . in section [ sec : scat ]
we present a method for treating the scattering of photons by material media . as an example of application of the proposed formalism , in section [ sec : cpd ] the state of the photon pairs generated by parametric down conversion
is obtained with an intuitive derivation . in section [ sec : conc ] we present our concluding remarks and summarize our results .
the bialynicki - birula
sipe wave equation for a photon described by the wave function @xmath1 is @xcite @xmath2 where @xmath3 is the planck s constant divided by @xmath4 and @xmath5 is an operator that acts on the photon helicity .
a spin one can be represented by a vector in the 3-dimensional complex space , so the vectorial character of @xmath6 incorporates its internal degrees of freedom .
the photon wave function can be decomposed in the eigenvectors of the helicity operator @xmath5 and also in its real and imaginary parts : @xmath7 if we associate the real vectors @xmath8 and @xmath9 with the electric and magnetic fields of the photon and @xmath10 and @xmath11 with the permittivity and the permeability of free space , together with the divergence - free condition of the photon wave function @xmath12 ( which is essential to arrive at the photon wave equation from einstein kinematics @xcite ) are completely equivalent to the set of the four maxwell equations in vacuum .
the eigenvalues of the helicity operator correspond to photons with left and right circular polarizations .
an arbitrary polarization state can be written as a particular combination of these two .
there is a peculiarity in the form for the photon wave function : its modulus squared has the unit of energy density , not of probability density , and agrees with the classical electromagnetic energy density .
this fact is associated with the problem of non - localization of the photon @xcite , and it happens that the energy density is the local useful concept , since photons have no other scalar quantities such as mass or charge , but only energy @xcite .
the normalization condition is that the integral of @xmath13 is equal to the expectation value of the photon energy .
there is no direct association of the modulus squared of the wave function with the probability density of detecting the photon in the region , unless we consider dimensions larger than the photon wavelength @xcite .
but neither the localization of massive relativistic particles is a clear concept . in this sense , photons are not significantly different from massive particles , as both can have an exponential falloff of their wave functions with distance in relativistic treatments @xcite , the difference being that photons are always relativistic .
our proposal to treat the interaction of photons with continuous media is to include a term proportional to the current density @xmath14 induced in the media due to the presence of the photon in : @xmath15 this approach is similar to the one adopted by keller @xcite .
however , keller was interested in calculating the wave function of a photon emitted from an atom and the current density was a source term .
here we are considering continuous media and the current density @xmath14 represents a response of the medium due to the presence of the photon , not being an independent source of photons .
so the objective of the present work is very different from keller s , although the initial step is similar .
when considering continuous media , we are implicitly assuming that the photon wavelength is much larger than the distances between the atoms that compose the medium , in such a way that the collective motion of the charges may be approximated by a continuous current density .
we can write @xmath16 where @xmath17 and @xmath18 are the magnetization and electric polarization of the medium and @xmath19 is the free current density , all of them being functions of the electric and magnetic fields of the photon .
for instance , in a non - dispersive dielectric medium the polarization can be written as @xmath20 expanded in linear and non - linear responses to the photon field . since the non - linear response terms represent situations in which there are more than one photon involved , a procedure of second quantization of the photon wave function is necessary to correctly describe them .
this issue will become clear in section [ sec : cpd ] where we treat the parametric down - conversion , which arises from the second - order non - linear response of the medium .
so , if we want to deal with the wave function of a single photon , we must disregard the non - linear response of the medium . in section [ sec : cpd ]
we will be interested in the interaction of photons with non - linear media after applying the second quantization procedure to the photon wave function , so we will continue referring to the current density @xmath14 in this section without specification of its linear or non - linear behaviors . using the photon wave function from , we can see that is equivalent to the maxwell equations by writing its real and imaginary parts .
its real part with the application of the operator @xmath5 and its imaginary part give : @xmath21 @xmath22 with @xmath8 , @xmath9 , @xmath23 and @xmath24 . taking the divergence of the above equations
we obtain @xmath25 where @xmath26 is the free charge density and we have @xmath27 .
what the above equations tell us is that if @xmath28 and @xmath29 at any particular instant of time , which is true before the photon interacts with the media , these relations must be always satisfied . with this argument
we complete the proof that is equivalent to the set of maxwell equations for electromagnetic waves .
so our treatment predicts that the behavior of photons in linear non - absorptive media , where a one - photon theory may be used , is completely analogous to the behavior of classical electromagnetic pulses .
this statement is supported by experiments @xcite and is difficult to obtain with the formalism of second quantization @xcite .
so we believe that in the same way that the schrdinger equation is used to treat the behavior of massive particles under the action of a classical potential , the formalism of this section may be used to describe the interaction of photons with material media when there is no absorption or emission of photons . in cases where there is absorption or emission of photons
we must apply the second quantization procedure on the photon wave function , to be done in the next section . from we
can derive a continuity equation for the photon wave function .
taking the scalar product with @xmath30 on both sides of and summing the complex conjugate , after verifying that @xmath31 and using the vectorial identity @xmath32 , we arrive at @xmath33 where @xmath34 is the current density of the photon wave function . using ( [ psi ] ) and taking the time average of @xmath35 over one period of oscillation of the photon field , we have @xmath36 , analogous to the electromagnetic energy flux .
the term @xmath37 represents the density of work done on the charges of the medium per unit time .
the continuity equation for the photon wave function is an energy continuity equation . in our formalism
, we are assuming that the photon energy density has always the same form when written in terms of the photon electric and magnetic fields , independently of the medium , and that the material energy density is calculated by the work done on the charges of the medium , without distinction whether the work is done on bound or free currents .
this approach differs from the one adopted by bialynicki - birula to treat photons in material media @xcite , where the photon wave function is written as @xmath38 and there is no current term in the photon wave equation , since the bound currents are included in the photon wave function via @xmath39 , @xmath40 and @xmath41 .
the photon energy density @xmath42 is a function of the electric permittivity @xmath43 and magnetic permeability @xmath44 of the medium since we have @xmath45 and @xmath46 in a linear isotropic medium . when a photon enters a medium , part of its energy is transferred to the medium .
the same occurs with its momentum .
the discussion of the correct form for the momentum and energy densities of classical electromagnetic waves in a medium has been the subject of a long debate @xcite .
the eventual conclusion is that there are many compatible ways of defining the electromagnetic and material parts of the energy and momentum densities of electromagnetic waves in material media .
here we are assuming a particular division in the quantum domain , where the electromagnetic parts are @xmath47 for the energy density , @xmath35 for the energy flux and @xmath48 for the momentum density .
this division has been successfully tested with reasonable models in the classical domain @xcite .
our formulation gives the same results as bialynicki - birula s for linear media : photons obey the maxwell equations , the difference being that we implicitly divided the total energy and momentum of the photon in the medium in their electromagnetic and material parts , associating only the electromagnetic part with the photon wave function .
as a consequence , our formulation can be readily extended to non - linear media after the procedure of second quantization of the photon wave function , representing an advantage to treat more general cases , as we discuss in the next sections .
the second quantization procedure can be applied to the one - photon theory to build a quantum field theory for describing states with an arbitrary number of photons .
this procedure is described in @xcite . here
we will make a similar second quantization procedure in an infinite linear non - dispersive and non - absorptive medium .
in such a medium , the material acquires an energy density @xmath49 $ ] due to the presence of the photon @xcite , @xmath40 and @xmath41 being the electric and magnetic susceptibilities of the medium , apart from the electromagnetic energy density @xmath47 .
so the total energy density of the photon in such a medium can be written as @xmath50 , where @xmath51 is the `` dressed photon '' wave function , with @xmath52 where @xmath53 and @xmath54 .
it is straightforward to show that @xmath55 .
this form for @xmath56 is equivalent to the one used by bialynicki - birula in the case of non - dispersive linear media @xcite .
however , if dispersion is included the modulus squared of the bialynicki - birula s form for the photon wave function in the medium does not give the total energy density anymore , while the division of the total energy density in electromagnetic and material parts remains valid @xcite .
although we will not treat dispersive media here , it is important to stress this fundamental difference between the treatments .
the above relations can be inverted , giving @xmath57\;.\end{aligned}\ ] ] the total energy of the `` dressed photon '' can be written as @xmath58 the `` dressed photon '' wave function can be decomposed in its energy eigenstates , which are plane - wave modes : @xmath59 with @xmath60 , where the plane - wave modes are defined by their wavevector @xmath61 and polarization vector @xmath62 , with frequency given by @xmath63 , @xmath64 being the refraction index of the medium . substituting ( [ psil r - p ] ) in ( [ en foton linear ] ) , the energy can be written as @xmath65 the above hamiltonian is equivalent to the hamiltonian of a continuous set of harmonic oscillators @xcite , and the result of the second quantization procedure is to raise the `` dressed photon '' wave function to the status of a field operator @xmath66 where @xmath67 is the annihilation operator for photons in the mode @xmath68 that , together with the creation operators @xmath69 , obey the commutation relations @xmath70=[\hat{a}_{\mathbf{k}s}^\dag,\hat{a}^\dag_{\mathbf{k}'s'}]=0,\;\ ; [ \hat{a}_{\mathbf{k}s},\hat{a}^\dag_{\mathbf{k}'s'}]=\delta^3(\mathbf{k}-\mathbf{k}')\delta_{ss'}.\ ] ] according to ( [ e - psil ] ) and ( [ b - psil ] ) , the electric and magnetic fields also become operators ) from ( [ b - psil ] ) , we must use the fact that @xmath71 , where @xmath72 is a polarization vector with helicity @xmath73 . this can be verified choosing @xmath74 , @xmath75 and @xmath76 . ] : @xmath77 @xmath78+\;\mathrm{h.c.}.\ ] ] equations ( [ op e ] ) and ( [ op b ] ) for the electric and magnetic field operators are identical to the ones obtained with the traditional second quantization of the electromagnetic field in vacuum if we substitute @xmath43 by @xmath10 and @xmath44 by @xmath11 @xcite . in the treatment of smith and raymer for the quantization of the photon wave function in vacuum @xcite , the operator @xmath79 they find is equivalent to our @xmath80 , where @xmath81 is given by ( [ op psil r - p ] ) with @xmath82 .
we believe that our form , in which @xmath81 has only annihilation operators , is more convenient , since it agrees with the condition @xmath83 derived from ( [ psil ] ) .
a one - photon state can be written as @xmath84 with @xmath85 and @xmath86 representing the vacuum state of the electromagnetic field .
the corresponding one - photon wave function is @xcite @xmath87 in agreement with ( [ psil r - p ] ) .
a two - photon state can be written as @xmath88 and the corresponding two - photon wave function is @xcite @xmath89\\ & & \times\sqrt{\hbar\omega}\mathrm{e}^{\rmi(\mathbf{k}\cdot \mathbf{r}_1-\omega t)}\mathbf{\hat{e}}_{\mathbf{k}s}\otimes \sqrt{\hbar\omega'}\mathrm{e}^{\rmi(\mathbf{k}'\cdot \mathbf{r}_2-\omega ' t)}\mathbf{\hat{e}}_{\mathbf{k}'s'}.\end{aligned}\ ] ] the tensor product in the above equation indicates the composition of two one - photon space states .
note that the wave function is automatically symmetrized via the term @xmath90 $ ] .
the extension to higher dimensions is straightforward .
here we want to find the behavior of the operators @xmath91 and @xmath92 from ( [ op e ] ) and ( [ op b ] ) when , besides the linear medium with electric and magnetic susceptibilities @xmath40 and @xmath41 occupying all space , there is another medium in a finite region with different properties . our treatment can be seen as an approximation of keller s treatment for the photon scattering @xcite in the far - field regime . applying the second quantization procedure including the interaction with matter , the operators @xmath79 will obey the same wave equation ( [ eq onda mat ] ) as the wave functions @xmath93 .
so the operators @xmath91 and @xmath92 will also obey ( [ max ph 1 ] ) and ( [ max ph 2 ] ) . equation ( [ max ph 2 ] ) can be written in the following way @xmath94 where @xmath95-\frac{\partial(\chi_\mathrm{e}{\mathbf{e}})}{\partial t}\ ] ] contain terms due to the difference of the behavior of the scattering medium in relation to the quantization medium , according to ( [ j clas ] ) . because the current density @xmath96 is a function of the electric and magnetic fields , when the fields become operators @xmath97
also becomes an operator .
differentiating ( [ max ph 2 b ] ) with respect to time , using ( [ max ph 1 ] ) and the vectorial identity @xmath98 , we obtain a wave equation for the operator @xmath91 : @xmath99 the above equation is a wave equation in which @xmath100 is the source .
an analog equation can be found for the operator @xmath92 in a similar way : @xmath101 solutions of ( [ eq onda source e ] ) can be obtained using green functions @xcite . if we want to study the scattering of a photon in a known state in the paraxial regime , where the dependence of the polarization with the propagation direction can be disregarded and a scalar theory may be used , the relevant solutions of ( [ eq onda source e ] ) for the operator @xmath102 are @xmath103 @xmath104 where @xmath105 . integrating ( [ g ] ) in @xmath106 , we see that the green functions can be written as @xmath107 it can also be shown , by a direct integration on the variables @xmath108 and @xmath109 in spherical coordinates @xmath110 @xcite , that @xmath111}=g^{(+)}(\mathbf{r},t;\mathbf{r}',t')-g^{(-)}(\mathbf{r},t;\mathbf{r}',t').\ ] ] if we are studying the scattering of a photon , the observation time @xmath112 will always be greater than the time of the interaction @xmath113 .
so , according to ( [ g delta ] ) , @xmath114 and we can write @xmath115}.\ ] ] the utilization of the green function method to predict the behavior of the electric field operator , according to ( [ psi out ] ) , is very useful for calculating the state of scattered photons by a localized material medium . in linear media
, we can use the same equation substituting the operator by the electric field of the photons , obtaining equivalent results to the scattering of an electromagnetic wave , since ( [ psi out ] ) is obtained from the maxwell equations .
but the interesting results come from the treatment of problems that can not be obtained from classical electrodynamics . in the next section
, we will apply the treatment to the scattering of a photon by a non - linear medium that can generate a photon pair in an entangled state .
we will show that the quantum state of the generated photon pair we obtain is the same as the one previously obtained @xcite , but with a more intuitive interpretation .
the parametric down conversion corresponds to the conversion of one incident photon in two by a non - linear birefringent crystal . as an example of application of the formalism developed in the previous section
, we will calculate the quantum state of the photon pair generated in the process of type - i parametric down conversion , in which both generated photons have polarization orthogonal to the incident photon . to simplify the treatment
, we will consider that the non - linear crystal is immersed in a medium with the same linear responses as the crystal .
so the medium where the field is quantized occupies all space , and the non - linearities of the crystal will be responsible for the scattering processes .
the magnetic responses of these crystals are very small , so we can consider a refraction index @xmath116 .
we will also disregard the dependence of the refraction index with the propagation direction and the effects of dispersion .
these approximations are consistent with the observation of photons at small angles in relation to the propagation direction of the incident photon and to the use of interference filters on the detectors , post - selecting states with relatively narrow frequency spectra . with all these considerations ,
the non - linear components of the polarization of the crystal can be written as @xmath117 where terms beyond the second order were disregarded . in type - i
parametric down conversion , the generated photons have ordinary polarization and the incident photon has extraordinary polarization .
according to ( [ j clas ] ) , the non - linear current density associated with the ordinary polarization can be written as @xmath118 , and the scattering term , according to ( [ eq onda source e ] ) , is @xmath119 .
so the relevant source term for the scattering in ( [ psi out ] ) is @xmath120,\ ] ] where the subscript @xmath121 ( @xmath122 ) denotes extraordinary ( ordinary ) polarization and we have @xmath123 , since @xmath124 and @xmath125 are operators that act on modes with orthogonal polarization and therefore commute .
according to ( [ funcao onda 2 fotons ] ) , if we denote the state of the incident photon by @xmath126 , the two - photon component with ordinary polarization of the output state is @xmath127 considering a first order scattering , in which one of the operators @xmath128 above is substituted via ( [ op psil - e o ] ) by the scattered electric field of ( [ psi out ] ) and the other is kept with the non - scattered field , we can write @xmath129 , that correctly symmetrizes the wave function . from now on we will omit this permutation term to simplify the notation .
the refraction index @xmath130 appears in the above equation because we have @xmath131 .
@xmath132 is given by ( [ psi out ] ) , with @xmath100 given by ( [ f cpd ] ) .
the state @xmath126 has one photon with extraordinary polarization and in ( [ psi 2 ] ) 2 operators that act on ordinary polarization modes are applied to the state @xmath126 .
we can substitute these operators by their commutator plus terms that result in zero when applied on @xmath126 . using ( [ op psil r - p ] ) ,
( [ rel comut ] ) and ( [ op e ] ) , the relevant commutator to the problem in the limit that we disregard the dependence of the polarization with the propagation direction is @xcite @xmath133= \int { \rmd^3k}\frac{\hbar\omega}{2n(2\pi)^3}\mathrm{e}^{\rmi[\mathbf{k}\cdot ( \mathbf{r}-\mathbf{r}')-\omega ( t - t')]}=\\ & & = \frac{-\rmi\hbar}{4c^2 } \frac{\partial^2}{\partial t \partial t'}g^{(+)}(\mathbf{r}_1,t;\mathbf{r}',t'),\end{aligned}\ ] ] where ( [ g2 ] ) was used in the last equality . inserting the value of ( [ f cpd ] ) for @xmath100 in @xmath134 , obtained from ( [ psi out ] ) , in ( [ psi 2 ] ) and using ( [ comut e ] ) to substitute the product of the operators @xmath135 and @xmath136 by their commutator plus terms that result in zero
when applied on @xmath126 , we have @xmath137e_3'(\mathbf{r}',t')\right\}\ ; , \end{aligned}\ ] ] with @xmath138 , we have @xmath139}\mathrm{e}^{\rmi[\mathbf{k}_2\cdot \mathbf{r}_1-\omega_2 t]}\;.\end{aligned}\ ] ] if we observe the generated photons at a distance from the crystal such that the time light takes to go from the crystal to the detectors is greater than the duration of the incident pulse , the integral in @xmath113 may be extended until @xmath140 , resulting in a term proportional to @xmath141 .
so the term @xmath142 in the above equation may be substituted by @xmath143 .
the above form for the wave function of the twin photons generated by parametric down conversion is equivalent to previous treatments that consider a hamiltonian evolution of the quantum state of light interacting with a non - linear medium using perturbation theory @xcite .
if we use the form from ( [ g ] ) to the green function , we arrive at a more intuitive form for the quantum state of the twin photons : @xmath144}}{|\mathbf{r}_1-\mathbf{r}'|}\omega_2 ^ 2\frac{\mathrm{e}^{\rmi[k_2|\mathbf{r}_2-\mathbf{r}'|-\omega_2(t - t')]}}{|\mathbf{r}_2-\mathbf{r}'| } \;.\end{aligned}\ ] ] we can interpret the above wave function as being the result of the coherent scattering of one photon in two by all non - linear scatterers of the crystal . in each one of these processes ,
the incident photon generates a pair of photons in spherical wave modes emitted from each non - linear scatterer . the energy and momentum conservation in the process
may be seen as consequences of the fundamental indistinguishability of all possible instants of time and positions in which a photon pair could be generated , which forces us to coherently add all the probability amplitudes .
these probability amplitudes interfere destructively when energy is not conserved .
however , although the sum of the energies of the twin photons is the energy of the incident photon , the energy of each photon can assume a wide range of values .
so , the photon pair is highly entangled in energy .
the probability amplitudes also interfere making the sum of the momenta of the twin photons be associated with the momentum of the incident photon , depending on the crystal geometry since the crystal may absorb part of the momentum in the process .
the momentum of each photon can also assume a wide range of values , so the generated photons are also entangled in linear momentum . comparing ( [ estado cpd ] ) and ( [ funcao onda 2 fotons ] ) , we see that the post - selected wave function of the twin photons on the wavevector space is proportional to @xmath145 let us consider that the incident photon is in a beam mode that propagates in the @xmath146 direction and that the crystal is a rectangular box with dimensions much larger than the beam waist in the @xmath147 and @xmath148 directions and a small width in the @xmath149 direction .
let us also treat the collinear regime , in which the crystal is oriented to maximize the production of photons with wavevectors near the @xmath146 direction .
so the integrals in @xmath147 and @xmath148 in ( [ 2photon wavenumber ] ) may be extended to infinity and the integral in @xmath149 results in a constant . to illustrate the energy and momentum entanglement of the state ,
let us represent the vectors @xmath61 as @xmath150 and @xmath151 as @xmath152 , since @xmath153 and @xmath154 determine @xmath155 .
let us also substitute the integral in the @xmath146 component of @xmath156 by an integral in @xmath153 . with all these approximations ,
the integral in @xmath113 gives @xmath157 and the integrals on @xmath158 and @xmath159 give @xmath160 .
so the 2-photon wave function on wavevector space can be written within the paraxial domain ( @xmath161 ) as @xcite @xmath162 we can not write this wave function as a product of a wave function for photon 1 and a wave function for photon 2 , what characterizes the entanglement of the state in both energy and momentum .
we extended the bialynicki - birula sipe photon wave function formalism to include the interaction between photons and continuous non - absorptive media .
the photon wave equation becomes equivalent to the macroscopic maxwell equations in material media . applying the second quantization procedure to the photon wave function
, we obtained an alternative treatment for the quantum interactions between light and continuous media , that can be specially useful to treat the scattering of light by matter .
as an example of application of the proposed formalism , the quantum state of the twin photons generated by parametric down conversion was obtained in agreement with previous treatments , but with a more intuitive interpretation .
the authors acknowledge brian j. smith for useful discussions .
this work was supported by the brazilian agencies cnpq and capes . | the bialynicki - birula sipe photon wave function formalism is extended to include the interaction between photons and continuous non - absorptive media .
when the second quantization of this formalism is introduced , a new way of describing the quantum interactions between light and matter emerges . as an example of application ,
the quantum state of the twin photons generated by parametric down conversion is obtained in agreement with previous treatments , but with a more intuitive interpretation . |
the diffusion equation is widely used to describe transport phenomena in which heat , mass , and other physical quantities are transferred in space and time due to underlying molecular collision processes @xcite . at steady - state and when the source term does not depend explicitly on the concentration of the associated field , such diffusion processes settle down in the form of a non - local relation between the spatial distribution of the source and the resulting concentration of the associated field , a relation which is governed by the poisson equation @xcite .
the notion of the poisson equation as the steady - state solution of the diffusion equation extends to many other phenomena , e.g. electrostatics systems @xcite , plasma physics @xcite , chemistry and biology @xcite , molecular biophysics @xcite , density functional theory and solid - state physics @xcite , to name but a few .
in fact , solving the poisson equation accurately and efficiently continues to be a subject of intense research to this day . from a microscopic point of view
, the diffusion equation emerges from the underlying boltzmann kinetic equation in the limit of very small knudsen numbers , i.e. molecular mean free path much shorter than the typical macroscopic scale .
solving the boltzmann kinetic equation to study diffusion processes makes apparently little sense , since the boltzmann distribution function lives in a double - dimensional phase - space , i.e. position and velocity .
however , in the last two decades , minimal ( lattice ) versions of the boltzmann equation have been developed , in which velocity space is reduced to a small set of discrete velocities , so that the computational cost is cut down dramatically , making the solution of the lattice kinetic equation often competitive towards standard grid - discretisation of the corresponding partial differential equation .
the lattice boltzmann method ( lbm ) has attracted much interest for solving the navier - stokes equations @xcite , and has been extended to solve other type of phenomena , e.g. diffusion equation @xcite , maxwell equations@xcite , quantum systems @xcite , relativistic fluid dynamics @xcite , and the poisson equation @xcite , to name but a few . to the best of our knowledge , to date , the solution of the poisson equation based on kinetic theory has been confined to smooth ( non - stiff ) source terms and small knudsen numbers @xcite .
if the target is the steady - state solution of the diffusion equation , as it is the case for the poisson equation , the kinetic approach looses much of its appeal , since its inherently real - time dynamics must proceed through smaller time - steps than those affordable by fictitious - time iteration techniques . in this paper
we shall show that such gap can be closed by resorting to higher - order kinetic formulations , at least for the cases where the source term does not depend on the concentration field avoiding artificial solutions as pointed out in ref .
@xcite .
higher - order formulations of the lb equation have been developed before , mostly in the context of thermal @xcite and relativistic fluid dynamics @xcite , and recently , density functional theory @xcite .
there , the main idea is to design the equilibrium distribution in such a way to include not only the conserved moments , such as density and current , but also the non - conserved ones , namely the momentum flux tensor and the heat flux , and eventually higher order kinetic moments with no direct hydrodynamic significance ( sometimes called `` ghosts '' ) .
clearly , in order to match this longer ladder of kinetic moments , a correspondingly larger set of discrete velocities is required .
it should be emphasized that the discrete kinetic equation is a superset of the diffusion equation , and consequently , to the purpose of solving the diffusion equation alone , higher - order contributions must be regarded as discretisation artefacts which need to be minimised , and possibly canceled altogether .
this is precisely the target of this paper . here
, we will use higher - order lattices and associated equilibria , to remove higher - order contributions to the macroscopic equations reproduced by the boltzmann equation . for this purpose
, we develop the respective lbm and show that , by expanding the equilibrium distribution up to seventh order in hermite polynomials , one can solve the poisson equation _ six orders _ of magnitude faster than with standard lb models , at a given level accuracy .
the paper is divided as follows : in sec .
[ sec1 ] , we introduce the diffusion equation with stiff source term and the associated boltzmann kinetic equation . in sec
. [ sec2 ] , we describe the influence of higher order moments on the steady - state solution of the diffusion equation . in sec .
[ sec3 ] , we develop a lbm for the poisson equation , validate and compare the model with standard cases . finally , we discuss the results and outlook of our work .
let us begin by writing the diffusion equation for a scalar @xmath0 in the standard form : @xmath1 where @xmath0 is the concentration of particles ( temperature in the case of the heat transfer equation ) , @xmath2 is the diffusivity , and @xmath3 is the source term , which , in our case , depending neither on @xmath0 nor on time .
the steady - state version of eq
. delivers the poisson equation with source @xmath4 .
it is well - known that this equation can be obtained from an underlying boltzmann kinetic equation , via a chapman - enskog expansion , in the limit of small knudsen numbers , @xmath5 , being @xmath6 the mean - free path and @xmath7 the characteristic size of the system @xcite . the boltzmann equation in the single relaxation time approximation ( bgk )
@xcite reads as follows : @xmath8 where @xmath9 is the probability distribution function , @xmath10 are the microscopic velocity vectors , @xmath11 is the relaxation time , and @xmath12 is the equilibrium distribution . in the non - relativistic context
, the local equilibrium is given by a maxwell - boltzmann distribution : @xmath13 where @xmath14 is the local flow speed .
such local equilibrium encodes the basic symmetries of newtonian mechanics , particularly galilean invariance , which is built - in via the dependence on the molecular speed relative to the fluid @xmath15 , rather than the absolute one , @xmath16 . in order to secure galilean invariance for _ any _ fluid speed , kinetic moments at _ all orders
_ should be matched on the lattice .
this would require an infinite connectivity , which is clearly unviable on any realistic lattice . as a result , lb local equilibria
are typically based on finite - order hermite expansions , of the form @xcite : @xmath17 where @xmath18 are the tensorial hermite polynomials of order @xmath19 , and @xmath20 is the weight function defined as , @xmath21 @xmath22 being the sound speed .
note that @xmath20 is the uniform _
equilibrium , corresponding the no - flow limit of the local equilibrium @xmath12 .
the coefficients @xmath23 ( kinetic moments ) are calculated by projecting the equilibrium distribution onto the respective hermite polynomial , @xmath24 the lowest - order kinetic moments have a direct macroscopic interpretation .
for instance , by integrating on the velocity space ( @xmath25 ) , @xmath26 where the last equality comes from the requirement of mass conservation on the collsion operator , namely : @xmath27 the diffusivity of the system is dictated by the relaxation time as @xmath28 it can be shown that the boltzmann equation recovers the diffusion equation in the limit of small knudsen numbers , which is related with the mean - free path and therefore with the relaxation time .
this is the strongly - interacting fluid regime , whereby the particle dynamics is collision - dominated , so that collective behaviour sets in on a short timescale @xmath11 .
the opposite limit of large knudsen numbers , hence large values of @xmath11 , corresponds to the free - particle motion ( ballistic regime ) , whereby memory of the initial conditions is kept for a very long time .
it is therefore clear that in the ballistic regime higher order moments of the equilibrium distribution do not relax and consequently the boltzmann equation does not converge to any standard diffusion equation . on the other hand , by increasing the relaxation time , hence the effective diffusivity , one would intuitively expect a quickest path to steady - state . in this respect
, it would be highly desirable to derive a diffusion equation in kinetic form , which can attain steady - state solution at relatively large knudsen numbers by exactly cancelling as many high - order terms as possible .
as discussed above , this can be achieved by zeroing as many higher order contributions as possible , while retaining the largest possible diffusion coefficient .
let us consider the case when all time derivatives are zero , @xmath29 . the boltzmann equation , eq .
, can be written as , @xmath30 by integrating this equation in velocity space , we obtain a steady - state continuity equation for the current density : @xmath31 where @xmath32 and @xmath33 .
multiplying eq . by @xmath10 and integrating again , we obtain @xmath34 with @xmath35 being the momentum - flux tensor and @xmath36 the source current .
the first term at the right - hand - side takes the form @xmath37 , where @xmath38 is the flow speed in the local equilibrium . for the case of a purely diffusive dynamics
this term is zero .
the second term contributes a shift @xmath39 to the flow speed , and it must also be set to zero for the case of pure diffusion .
finally , the pressure tensor shoud reduce to @xmath40 , so that inserting ( [ second ] ) into ( [ first ] ) , the diffusion equation is obtained .
the first two conditions are automatically ensured by setting @xmath41 in the local equilibrium .
the third one , however , can not be enforced exactly because of higher order contributions to the momemtum flux tensors .
indeed , by iterating the procedure , one obtains the general expression : @xmath42 = s \quad , \ ] ] where @xmath43 and @xmath44 , are both tensors of order @xmath19 .
note that for small knudsen numbers ( @xmath45 ) higher - order terms become negligible and the series is convergent .
we also observe that the solution is dictated by the equilibrium distribution and the source term , so that , by properly choosing both expressions , higher order terms can be canceled thus opening the way to high - accuracy solutions of the poisson equation . in numerical practice , accuracy is typically improved by increasing the resolution of the grid .
this imposes correspondingly smaller time steps , which is clearly expensive , especially in three dimensions .
high - order methods are meant to mitigate the problem by achieving the same level of accuracy with many less grid points .
they are typically based on higher - order stencils for the discretisation of the corresponding differential operators . in this work
, we present a procedure whereby the same goal is achieved by drawing directly from an underlying lattice kinetic theory , i.e. by a suitable ( non - gaussian ) extension of the local kinetic equilibria as combined with the use of larger sets of discrete speeds than those typically employed in standard lattice boltzmann theory .
in order for the boltzmann kinetics to converge to the poisson equation , the following conditions need to be met : @xmath46 the equilibrium distribution can be expressed as a separable product of a function of the microscopic velocity ( global uniform equilibrium ) and a function of the time - spatial coordinates , namely : @xmath47 and similarly for the source term : @xmath48 in the above , @xmath49 and @xmath50 are velocity dependent functions that need to be determined based on the conditions above . to this purpose ,
we write : @xmath51 and , @xmath52 where @xmath53 and @xmath54 are constant coefficients and @xmath55 is a cut - off truncating the hermite expansion . with reference to the one - dimensional case and an expansion up to seventh order ( @xmath56 )
, we obtain : @xmath57 and @xmath58 + \frac{3c_s^2 - 6 c_s^2 v^2
+ v^4}{8 c_s^4 } + \frac{15 c_s^6 - 45 c_s^4 v^2 + 15 c_s^2 v^4 - v^6}{48 c_s^6}\right ) , \ ] ] where the weight @xmath20 for the one - dimensional case is defined as @xmath59 with eqs . and
, one can prove that the the following expressions for the moments are fulfilled : @xmath60 , @xmath61 , @xmath62 , @xmath63 , @xmath64 , @xmath65 .
these expressions hold only up to @xmath66 .
to develop a lattice boltzmann model and solve the poisson equation numerically , we need to impose a quadrature and find a corresponding set of finite velocity vectors , @xmath67 , such that the orthogonality conditions are fulfilled to the desired order @xcite , namely : @xmath68 here @xmath69 are discrete weights and @xmath70 are the one - dimensional hermite polynomials .
the number of velocity vectors @xmath71 depends on the order of the expansion that we intend to reproduce with the quadrature . for expansions up to seventh order , we need at least @xmath72 velocity vectors and weights .
each set of numbers also provides its own lattice sound speed @xmath22 , which goes from about @xmath73 with @xmath74 to nearly @xmath75 with @xmath56 .
the details of these values are given in [ app ] .
higher - order schemes are usually exposed to numerical instabilities , due to the appearance of additional modes in the dispersion relation . however , since our procedure is designed to annihilate precisely these modes , we expect our approach to be stable . furthermore , the method proposed here can also be used as a systematic technique of calculating weights and velocity vectors , such that one can achieve the desired level of accuracy for the computation of laplacian operators , similar to the work proposed in refs .
@xcite .
we have performed the theoretical analysis in one - dimension , however , extensions to two- and three - dimensions are straightforward , by using the tensorial form of the hermite polynomials and by finding the discrete velocity vectors with the respective orthogonality conditions .
for instance , in order to fulfill the conditions , up to fifth order , the following expansions apply : @xmath76 and @xmath77 + \frac{15 c_s^4 - 10 c_s^2 \vec{v}^2 + \vec{v}^4}{8 c_s^4 } \right ) , \ ] ] where the weight @xmath78 is defined as @xmath79 for the three - dimensional case , we should calculate the corresponding set of discrete velocity vectors , @xmath80 , such that the orthogonality condition is fulfilled up to fifth order , namely : @xmath81 which is the three - dimensional version of eq . .
by solving these algebraic equations , we find that we need at least @xmath82 velocity vectors and @xmath83 weights .
details are given in [ app ] .
as an application , we solve the poisson equation @xcite , @xmath84 where @xmath85 is the charge density and @xmath86 is the electric permittivity . replacing the equilibrium distribution @xmath12 from eq . into eq .
, we obtain : @xmath87 implying that @xmath88 . in the lattice boltzmann model
, one must take into account second order corrections to the relaxation time , @xmath89 , where @xmath90 the lattice boltzmann relaxation time . from eq .
, it is appreciated that the spatial derivatives of the source also introduce errors in the solution .
thus , by annihilating higher order contributions , we are also eliminating spurious effects due to the derivatives of the source term .
if the source term is smooth , this derivatives are negligible , however , for stiffer source terms , one can use the present approach to add accuracy to the solution . in order to investigate this issue , we solve the potential for the following charge density : @xmath91 where the integer @xmath92 controls the smoothness of the derivatives of the charge density , and @xmath7 is the simulated length , with @xmath93 .
we use periodic boundary conditions for simplicity , and @xmath94 ( numerical units ) . and @xmath95 .
the absolute error is computed @xmath96 , where @xmath97 is the potential calculated with the lattice boltzmann model , and @xmath98 is the analytical solution of eq . .
the system size is @xmath99 lattice cells.,title="fig : " ] and @xmath95 .
the absolute error is computed @xmath96 , where @xmath97 is the potential calculated with the lattice boltzmann model , and @xmath98 is the analytical solution of eq . .
the system size is @xmath99 lattice cells.,title="fig : " ] and @xmath100 .
the absolute error is computed @xmath96 , where @xmath97 is the potential calculated with the lattice boltzmann model , and @xmath98 is the analytical solution of eq . .
the system size is @xmath99 lattice cells.,title="fig : " ] and @xmath100 .
the absolute error is computed @xmath96 , where @xmath97 is the potential calculated with the lattice boltzmann model , and @xmath98 is the analytical solution of eq . .
the system size is @xmath99 lattice cells.,title="fig : " ] and @xmath100 .
the absolute error is computed @xmath96 , where @xmath97 is the potential calculated with the lattice boltzmann model , and @xmath98 is the analytical solution of eq . .
the system size is @xmath101 lattice cells.,title="fig : " ] and @xmath100 .
the absolute error is computed @xmath96 , where @xmath97 is the potential calculated with the lattice boltzmann model , and @xmath98 is the analytical solution of eq . .
the system size is @xmath101 lattice cells.,title="fig : " ] we express the relative error as @xmath102 where the amplitude @xmath103 and scaling exponent @xmath104 ( order of accuracy ) both depend on @xmath55 . from fig .
[ fig1 ] , we see that for a relatively large system size , @xmath105 , the first order expansion is sufficient to produce satisfactory results , with errors around @xmath106 for @xmath107 and @xmath74 , and below @xmath108 for @xmath109 and @xmath56 . in fig .
[ fig2 ] , we see that by increasing the derivatives of the charge density ( increasing @xmath92 ) , the error increases for all values of the cut - off @xmath55 .
however , as we have mentioned before , the effect of removing higher - order errors becomes crucial for the case of small system sizes . by decreasing the number of lattice cells ( see fig .
[ fig3 ] ) , the discrepancies become visual and the errors are around @xmath110 for @xmath111 , @xmath112 for @xmath113 , and less than @xmath114 for @xmath66 . and @xmath95 .
the absolute error is computed @xmath96 , where @xmath97 is the potential calculated with the lattice boltzmann model , and @xmath98 is the analytical solution of eq . .
the system size is @xmath115 lattice cells.,title="fig : " ] and @xmath95 .
the absolute error is computed @xmath96 , where @xmath97 is the potential calculated with the lattice boltzmann model , and @xmath98 is the analytical solution of eq . .
the system size is @xmath115 lattice cells.,title="fig : " ] finally , by further reducing the lattice resolution , @xmath115 cells , and considering only one oscillation , @xmath95 , we see that the errors for the first order expansion , @xmath107 , are around @xmath116 , while for the highest values of @xmath66 they remain below @xmath117 .
the fact that one can obtain very small errors by increasing the order of the expansion in the velocity space of the equilibrium distribution and source term , opens up the possibility of highly accurate solutions on very small grids . increasing the order of the expansion inevitably implies an overhead of numerical operations per time step , thereby slowing down the simulations . on the other hand , higher order lattices also bring aditional side - benefits such as a higher sound speed @xmath22 , leading to a larger diffusion coefficient and , consequently , to a faster path to the steady - state solution . in order to highlight the concrete advantages of our approach
, we next implement the same simulations for different orders @xmath55 and inspect the cpu time , number of iterations , and system size , required to achieve a specific level of accuracy .
let us next change the order of the expansion and analyse how the relative error decreases with the system size .
for each cut - off value @xmath55 of the hermite polynomials expansion .
the lines denote the fitting curves using the general expression , @xmath118 ( right ) .
number of iterations as a function of the system size @xmath7 for each cut - off value of @xmath55 .
the lines denote the fitting curve using the general expression , @xmath119 , @xmath120 being the number of time - steps.,title="fig : " ] for each cut - off value @xmath55 of the hermite polynomials expansion .
the lines denote the fitting curves using the general expression , @xmath118 ( right ) .
number of iterations as a function of the system size @xmath7 for each cut - off value of @xmath55 .
the lines denote the fitting curve using the general expression , @xmath119 , @xmath120 being the number of time - steps.,title="fig : " ] from fig .
[ fig5 ] ( left ) , we see that , according to our expectations , by increasing the order of the expansion , the order of convergence , measured by the exponent @xmath104 , also increases and in a very substantial way . from the same figure ( right )
, we can also observe that for all values of @xmath55 , the number of time - steps @xmath120 scales quadratically with the lattice size , and decreases with the order of the expansion @xmath55 .
the former feature is the typical signature of diffusion process , while the second reflects the fact that higher order lattice deliver a larger sound speed , hence they take less time - steps to achieve a given level of accuracy .
the computational time required to achieve a given accuracy clearly depends on the computational speed of the model on each given machine .
this is conveniently expressed in terms of the number of lattice sites updates per cpu second , @xmath121 .
this quantity is expected to be a decreasing function of @xmath55 , since higher - order quadratures require more operations per lattice site .
actual measurements give the values reported in table [ table1 ] .
although no clear trend can be extracted from these measurements , it is observed that the cases @xmath109 and @xmath56 are about five times more expensive than the cases @xmath107 and @xmath74 . also to be noted that @xmath122 , which is counter - intuitive .
however , upon inspecting the velocity vectors ( see appendix [ app ] ) , one can appreciate that for @xmath56 the velocities are consecutive ( no velocity gap ) , while for @xmath109 they are not .
since , memory access is faster for consecutive elements , we expect the case @xmath56 to perform better than @xmath113 .
more generally , the dependence on @xmath55 appears to be highly dependent on memory access patterns , which are not easily quantified through simple scaling relations .
the total cpu time , @xmath123 , it takes to attain steady - state for a given system size @xmath7 can be written as follows : @xmath124 in the above , the cubic dependence derives from the @xmath125 diffusive scaling , as combined with the number of sites @xmath7 ( the space - time computational volume of a diffusive process scales like @xmath126 in @xmath2 spatial dimensions ) .
finally , @xmath127 is a relative scale for the number of time - steps , which is expected to decrease at increasing @xmath55 because of the increasing sound speed .
the actual values of obtained in the simulations are reported in table [ table1 ] . as an example , for the case @xmath128 , we find that the simulations take @xmath129 ( 65629 time steps ) , @xmath130 ( 65629 time steps ) , @xmath131 ( 33769 time steps ) , and @xmath132 ( 20079 time steps ) seconds , for @xmath133 respectively .
this shows that the total computational time remains within a factor two across all values of @xmath55 , while the error with respect to the analytical solution decreases dramatically at increasing @xmath55 ( see fig .
[ fig5 ] ) .
hence , we can conclude that increasing @xmath55 leads to a dramatic boost of accuracy at a very moderate extra computational cost .
finally , we have inspected the computational time to steady - state for a given accuracy , at changing the size of the problem and the order of the quadrature . by combining the previous relations , ( [ accur ] ) and ( [ tl3 ] ) ,
we obtain : @xmath134 the plot reports the computational time and system size needed to achieve a prescribed relative error , for the case in point @xmath135 .
and system size @xmath7 needed to recover the solution of the poisson equation with a relative error of @xmath136 , for different @xmath55 . ] from fig .
[ fig6 ] , we see that by increasing the order of the expansion in hermite polynomials , it is possible to reduce the computational time by several orders of magnitude ( up to six ) .
these data refer specifically to the solution of the poisson equation , but we have reasons to believe that similar conclusions would hold for other partial differential equations with a kinetic - theory background , including many linear non - linear partial differential equations of great relevance in many branches of modern science . in order to gain information about the performance of this approach , as compared with existing tools to solve the poisson equation , we have implemented simulations using the multi - grid ( mg ) method @xcite . the number of nodes in the mg simulations is denoted by @xmath7 , like for the lattice boltzmann method .
fixing @xmath95 ( from eq . ) , in table [ table2 ] we show the relative errors per node , @xmath137 , together with the corresponding computational time .
note that for @xmath74 , the mg method is faster than our approach , with the same second order accuracy . by increasing @xmath55 , however , the mg becomes increasingly faster , but also increasingly less accurate at a given grid size @xmath7 .
for instance , in order to attain the same accuracy of the case @xmath56 on @xmath138 , the mg solver requires @xmath139 grid points , being still faster by about a factor @xmath140 .
given that @xmath56 requires @xmath72 arrays , the lb memory saving is about a factor @xmath141 .
these figures indicate that the present high - order lb poisson solver is consistently slower than mg , however it also takes correspondingly less memory to achieve high levels of accuracy .
given that mg represents the state - of - the art for fast poisson solvers , these can be regarded as satisfactory results .
in summary , we have shown that the present lb approach provides a viable option for the solution of poisson equation , in the sense of simplicity and possibly also in terms of memory demand at high accuracy .
summarizing , we have studied the effect of higher order terms on the steady - state solution of the diffusion equation , using a lattice version of the boltzmann equation with high - order ( non - gaussian ) equilibria . for the numerical solution
, we have developed a higher order lattice boltzmann model capable of reproducing up to seventh order moment of the equilibrium distribution and source charge .
this permits to cancel exactly error terms up to seventh order . for the validation and study of our approach
, we have solved the one - dimensional poisson equation , and found that the computational time to steady - state , at a prescribed level of accuracy , can be reduced by up to six orders of magnitude as compared to standard lb formulations .
the six orders gain in computational time is due to the fact that higher order lattices permit to utilize much smaller grid sizes .
moreover , since they contain a large number of discrete velocities , they allow faster propagation within the lattice , hence a correspondingly faster attainement of the steady - state . we have also compared with state - of - the art multigrid solvers , and found that high - order lb is nearly competitive in efficiency at a lower memory demand .
extensions to two and three - dimensions , as well as time - dependent diffusion and advection - diffusion equations will make the object of future research .
we acknowledge financial support from the european research council ( erc ) advanced grant 319968-flowccs .
using eq . for each order of expansion @xmath55 , we can calculate the velocity vectors @xmath67 with the respective weights @xmath69 .
in addition , each set of @xmath69 and @xmath67 provides a characteristic normalised speed @xmath22 , @xmath142 .
here we only show the cases for @xmath143 , and @xmath144 , since the simulations for @xmath145 were performed with the same lattice than for @xmath146 . for @xmath146
, we have @xmath147 for @xmath113 , @xmath148 and finally , for @xmath66 , @xmath149 | we present a new approach to find accurate solutions to the poisson equation , as obtained from the steady - state limit of a diffusion equation with strong source terms . for this purpose ,
we start from boltzmann s kinetic theory and investigate the influence of higher order terms on the resulting macroscopic equations . by performing an appropriate expansion of the equilibrium distribution ,
we provide a method to remove the unnecessary terms up to a desired order and show that it is possible to find , with high level of accuracy , the steady - state solution of the diffusion equation for sizeable knudsen numbers . in order to test our kinetic approach ,
we discretise the boltzmann equation and solve the poisson equation , spending up to six order of magnitude less computational time for a given precision than standard lattice boltzmann methods . |
the advanced ligo experiment detected the first transient gravitational wave signal ( gw150914 ) from the inspiral and merger of a pair of black holes of masses 36 and 29 @xcite .
this was remarkable not only for being the first direct detection of gravitational waves but the first evidence that binary black holes ( bbh ) exist , and the largest mass estimates for black holes in the stellar regime @xcite .
this has been followed by a second discovery , also of a bbh merger signal , with a pair of bhs with masses @xmath5 and @xmath6@xcite on 2015 december 26 ( gw151226 ) .
ligo estimate a luminosity distance of @xmath7mpc corresponding to a redshift @xmath8 ( 90% limits ) . ] a broad range of teams have begun efforts to follow - up gw signals to detect the putative electromagnetic ( em ) counterparts .
the first event resulted in 25 teams of observers covering the ligo sky localization region with gamma ray to radio facilities ( summarised in * ? ? ? * ) .
the general assumption has been that bbh mergers will not produce a detectable em signature .
however fermi may have detected a weak x - ray transient which was temporally coincident with gw150914 @xcite , although the reality of the detection is disputed by @xcite .
@xcite suggested a novel mechanism that may produce both a bbh merger and a relativistic jet from the fragmentation of a rapidly rotating core of a single massive star .
however if the fermi hard x - ray detection is real , it is more like a short gamma - ray burst than a long one .
furthermore @xcite investigated this scenario quantitatively and finds a single star origin to be unlikely .
@xcite proposed a short grb may be formed if the two black holes are formed within a fossil disk which restarts accretion due to tidal forces and shocks during the bbh merger .
hence the searches continue , particularly as a detection would open up a major new way to probe high energy astrophysics , stellar evolution , compact remnants and test modified theories of gravity @xcite . here
we present the results of our wide - field search for an optical counterpart to the transient gravitational wave event gw151226 using the pan - starrs1 ( ps1 ) and the atlas survey telescopes combined with spectroscopic follow - up from hawaiian facilities and the public eso spectroscopic survey of transient objects ( pessto ) .
to search for optical counterparts to gravitational wave events our collaboration @xcite uses the pan - starrs1 system @xcite for imaging and relies on the existence of the pan - starrs1 3@xmath9 survey ( chambers et .
2016 in prep ) for template images .
the pessto survey together with programs on gemini north with gmos , the uh2.2 m with snifs , provide spectroscopic classification .
the data for one object discovered here were supplemented with hubble space telescope observations .
gw151226 was detected on 2015 december 26 03:39 utc ( mjd 57382.152 ) and released to the em community as a discovery on 2015 december 27 17:40 utc @xcite .
the initial localization generated by the bayestar pipeline @xcite contained a 50% credible region of 430 square degrees and a 90% region of about 1400 square degrees ( to be compared with 90% credible region of 630 square degrees for gw150914 * ? ? ? * ) .
we began taking data with the pan - starrs1 telescope during the next available dark hours , on 2015 dec 28 05:08 utc ( 11.47hr after the ligo information release and 49.48hr after the event time ) and mapped out a region of 214 square degrees on this first night as shown in figure[fig : skymap ] .
the same region was mapped on the two subsequent nights ( extending to 273 square degrees ) .
all observations were done with the pan - starrs1 @xmath0 filter with a 4-point dither pattern at each pointing centre .
the four individual ( back to back ) 45 sec exposures were co - added to produce a 180s exposure and the pan - starrs1 3@xmath9 @xmath0reference image ( typically having an effective total exposure time of 270 - 900 sec ) was subtracted from this 180s night stack ( see * ? ? ?
* and magnier et al . in prep , for more details ) .
on any one night this 180s exposure sequence was repeated multiple times ( 2 - 3 ) in the central highest probability region , giving us some intra - night sampling .
the sequence was repeated a further 5 times between 2016 january 02 and january 25 ( extending the full footprint to a total of 290 square degrees ) . the observing cadence and sensitivity
are illustrated in figure[fig : skymap ] , and the full ps1 footprint corresponds to 26.5% of the full ligo posterior probability .
this footprint choice was a combination of telescope accessibility of the ligo localization map and a choice to go deeper on the higher probability regions @xcite .
we selected targets with similar filtering algorithms as described in our first paper @xcite .
a total of 2.3@xmath10 detections were ingested into the database ( after basic rejections of known defects ) .
spatial aggregation of detections within 05 of each other resulted in the creation of 1.1@xmath10 objects and basic filtering and insistence of two separate detections resulted in a total of 1.7@xmath11 candidate astrophysical transients .
subsequent filtering ( obvious dipoles , stellar objects and objects near bright stars ) and a random forest machine learning classifier reduced the numbers to 144,000 for which the pixel recognition machine learning technique was employed @xcite .
further removal of 3,903 known minor planets left a total of 24,100 objects for humans to scan and this manual process resulted in 85 objects for further investigation .
the human scanning involved removing artefacts that are obvious to the eye but are not properly recognised by the machine learning .
as we wanted to err on the side of completion over purity , we set the machine learning threshold to roughly a 20 per cent false positive rate on the roc curve ( see fig.7 of * ? ? ?
* for an illustration ) .
the human scanning removed subtraction and chip defects that are easily distinguished visually .
we note that in the milky way plane there were at least a further 43 faint transients which are very likely variable stars that reach above our detection limit on a few epochs .
a few could be background hostless supernovae , but their location in the plane suggests they are faint stellar variables .
in addition , the atlas 0.5 m telescope @xcite , covered a significant fraction of the northern sky in the first five days after gw151226 as shown in figure[fig : skymap ] .
these data were taken during normal atlas operations and can be thought of as atlas working in serendipitous mode . in future , atlas will be able to work in targeted mode in the same way as pan - starrs1 .
a single atlas unit , with its 30 square degree cameras can map out 1000 square degrees within 30 minutes .
we highlight that just 3hrs after the gw151226 event detection , atlas serendipitously covered 87 square degrees of the sky localization region ( 2.2% enclosed probability ) during the time window @xmath12 .
we processed all atlas data taken serendipitously in the first 5 days to locate transients as in @xcite . after processing about 575 sq .
degrees , the atlas coverage increases the total enclosed probability to 36% over the first 5 days from gw151226 , getting to median 5@xmath13 limits of @xmath14 ( orange filter ) .
apart from variable stars and cv candidates , we found no other extragalactic transient candidates in this stream .
filters only ( see * ? ? ?
* for details ) and a full bolometric lightcurve estimated from a black body extrapolation between @xmath15 m . a simple arnett model , as described in @xcite is shown for the latter with input parameters : @xmath16ergs , @xmath17 , @xmath18 .
this simply indicates that a @xmath19ni model is a poor fit and that type ibn are not well explained by radioactive powering .
* lower : * comparison with two well observed sn ibn @xcite . ] during our filtering , we removed obvious galactic stellar variables and known agn candidates . after removing these contaminants
, we found 49 transients which are either confirmed sne or likely sne which are all summarised in table[tab : trans ] .
as discussed in @xcite the detected transients are dominated by mostly old supernovae that exploded over an extended period before the gw trigger .
the sky position of transients found in the first three days are plotted in figure[fig : skymap ] .
those with spectroscopic classifications are listed along with their redshifts .
we suggest that all these objects are unrelated field supernovae , although one object , ps15dpn deserves closer inspection and is discussed in the next section .
we note two objects that are unrelated to gw151226 but are worth highlighting in the context of searching for unusual transients in ligo / virgo sky localization regions .
ps15dqa is a faint transient in the nearby ( @xmath20mpc ) galaxy ngc 1156 .
the transient magnitude @xmath0=20.8 implies @xmath21 ( including significant milky way foreground extinction of @xmath22 ) . a hubble space telescope ( hst ) archive image with the advanced camera for surveys ( acs , in filter f625w ) of ngc 1156 shows an object which is astrometrically coincident with ps15dqa to within 03 .
this is a stellar point source with @xmath23 and hence @xmath24 . assuming a bolometric correction of zero this corresponds to @xmath25dex which implies a 50 - 60 star that is undergoing brightness variations by a factor of 2 .
while this scale of variability is known for massive stars of this luminosity , and is unrelated to gw151226 , it illustrates our ability to identify faint transients in nearby galaxies . secondly , ps16li is a fast optical transient with a 1.3 magnitude fade in 13min on the night of mjd=57397.51 with a faint , red point source in the ps1 reference stack
this is an m - dwarf flare ( e.g. * ? ? ?
* ) which highlights our ability to pick up fast decaying transients .
images from ps1 on mjd=57386 showing ps15dpn offset from its host galaxy .
color composite ( f475w , f814w , f160w ) image from hst on mjd=57447 . ]
this object was reported early in the campaign as being of interest because of its rising light curve and very blue spectrum @xcite .
we gathered a multi - color ps1 lightcurve in @xmath26 @xcite and one epoch of hst imaging with wfc3 ( see figures[fig:15dpn - lc ] and [ fig:15dpn - hst ] ) , together with 8 epochs of spectra .
the redshift of the host galaxy is measured at @xmath3 ( @xmath27mpc ) from the centroids of the strong host galaxy emission lines of h@xmath28 , [ nii ] and [ sii ] .
figure[fig:15dpn - spec ] shows the evolution of this transient into a type ibn supernova .
these sne are likely the explosion of wolf - rayet stars which are embedded in a he - rich circumstellar medium lost by the progenitor system @xcite .
the gmos spectrum at + 26 days post - peak is typical of this class with hei emission lines .
the hei @xmath295876 line has fwhm@xmath30 . to estimate the explosion epoch
, we used a third order polynomial fit to the first six ps1 @xmath0-band detections to estimate a date of @xmath31 .
this is @xmath32d before the detection of gw151226 .
the uncertainty is estimated from usage of different fits ( order 2 - 4 ) and epochs ( 4 to 8) . using the first 6 epochs and a 4th order fit gives an explosion epoch of 57382.03 , exactly coincident with gw151226 .
there is one type ibn which has a double peaked lightcurve @xcite and if that were common then this method would not be accurate .
we calculated the bolometric lightcurve after applying suitable @xmath33corrections @xcite . a comparison with the only ibn with an early discovery and well measured rise ( sn 2010al ; * ? ? ?
* ) , shows that the two have very different lightcurve shapes , and therefore sn 2010al is not a good template to use for dating the explosion .
this lightcurve diversity is a feature of ibn , likely indicating the diverse masses of the csm and ejecta that power these transients @xcite .
this illustrates that ps15dpn was temporally coincident with gw151226 to within 2.45 days .
another unusual feature of ps15dpn is the detection in the radio by the vla by @xcite , which is quite a luminous 6ghz detection , similar to the relativistic sn 2009bb @xcite .
therefore ps15dpn caught our attention because of its rarity , and also the fact that the remarkable pre - explosion outburst found for the nearest ibn ( sn2006jc * ? ? ?
* ) is still quantitatively unexplained . given the temporal coincidence of ps15dpn with gw151226 , we estimate the probability of finding a sn ibn randomly in our sampled field .
while we detected ps15dpn at @xmath34 , we would be sensitive to such an object with @xmath35 ( restframe absolute magnitude ) to @xmath36 . following the calculations in our first paper ( section 6.3 , * ? ? ?
* ) , the cosmic rate of core - collapse sne within @xmath36 implies that within 100 square degrees , there should be 3.1 ccsn explosions per day .
we assume an uncertainty in the explosion epoch estimate of ps15dpn of @xmath37 days and a relative rate of ibn sne of @xmath38 ( which is the fraction of core - collapse sne that are ibn ) .
then the number of ibn sne within a survey area of @xmath39 square degrees is expected to be @xmath40 for @xmath41 , @xmath42 , @xmath43 , this suggests @xmath44 .
hence the probability of a false positive ( 1 or more events ) when the expectation value is 0.18 is simply the poissonian value @xmath45 . in other words
the probability of finding an unrelated sn ibn exploding within 2 days of gw151226 which is unrelated and a chance coincidence is @xmath46 .
this is not convincingly low enough to imply a causal link , but is low enough to highlight that future coincidences should be searched for .
one could argue that the appropriate value to use for @xmath38 is significantly less than 0.01 , since radio detections at this luminosity for any ibc sn are quite rare @xcite .
alternatively , type ibc sne are much more common overall , which would increase the value for @xmath38 significantly .
one could speculate that a sn ibn could potentially be related to a gw source if it was a compact wolf - rayet star + bh binary , such as the wr+bh systems in the nearby galaxies ic10 and ngc300 @xcite . the wr star would need to undergo core - collapse supernova , followed by gravitational - wave driven merger with the bh companion within @xmath47 days .
however , assuming that the star was not in contact with the black hole prior to the supernova , the merger would typically require thousands of years rather than days .
a very favourable supernova kick toward the bh could reduce the merger timescale , but would require an implausibly high kick velocity and/or a very low - probability kick direction .
however the strongest argument against a link is that the distance estimate to gw151226 is inconsistent with the redshift of ps15dpn @xcite the final probability density function from ligo drops to zero at @xmath34 ( @xmath27mpc ) as shown in the detailed companion analysis paper @xcite .
while the mechanism of @xcite might predict a rapidly rotating massive star which could conceivably produce both a sn ibn and gw emission this now seems unlikely from the calculations of @xcite .
some luminous supernovae have been explained by magnetic neutron stars born with millisecond periods @xcite , and such an object would radiate gravitational waves if it were elliptically deformed .
however a neutron star origin is excluded as the ligo analysis is not consistent with component masses less than 4.5(99% credible level ; * ? ? ?
assuming that none of the transients we found , including ps15dpn , are associated with gw151226 it is useful to set quantitative and meaningful upper limits on potential optical counterparts for bh - bh mergers .
these also serve as a guide to our sensitivity to potential future binary neutron star ( ns - ns ) and neutron star
black hole ( ns - bh ) merging systems that are more promising systems for producing electromagnetic counterparts particularly redwards of 7000 . in figure1
we show the 5@xmath13 limits of every ps1 image taken during this campaign .
as described in @xcite , the 5@xmath13 limits are calculated for each of the 51 skycells per pointing of the gpc1 camera data products and the median per night is also plotted .
we also plot parametrised lightcurves of three analytic lightcurves with timescales @xmath48d ( as defined in * ? ? ? * ) .
these indicate detection limits of @xmath0=20.3 , 20.8 and 20.8 respectively , or @xmath49 at the luminosity distance of gw151226 . looking to the future , we plot model lightcurves of kilonovae from compact binary mergers ( ns - ns ) of @xcite and @xcite as illustrative examples of our survey capability ( the merger models of * ? ? ?
* ; * ? ? ?
* are also of similar luminosity ) . at the estimated distance of gw151226 of @xmath50mpc
, the predicted fluxes would be very faint ( below @xmath0@xmath51 ) .
it is expected that ns - ns mergers will be more common by volume and ligo s horizon distance for ns - ns detections is a factor of @xmath525 - 10 smaller than for bh - bh mergers , depending on the bh masses @xcite . during
the next science run beginning in the fall of 2016 , ligo is expected to be sensitive to ns - ns mergers within @xmath53mpc @xcite and we show in figure[fig : skymap ] that our survey strategy would be sensitive to these . ideally , our goal would be to get 0.5 mag deeper , beginning within 24hrs of the gw alert .
we further show the @xmath54band lightcurve of sn 1998bw @xcite which is the typical energetic type ic sn associated with long duration gamma - ray bursts ( lgrbs ) .
figure1 shows that if an energetic type ic sn accompanies such a grb event then it would be an unambiguous , bright transient in our survey .
we do not find such an object , but caution that we surveyed a maximum of 26.5% of the total ligo probability region .
our results are encouraging for future searches for the counterparts of ns - ns mergers within about 100mpc where the predicted optical and near infra - red counterparts are within reach .
ps1 and atlas are supported by nasa grants nnx08ar22 g , nnx12ar65 g , nnx14am74 g and nnx12ar55 g . based on data from : eso as part of pessto ( 188.d-3003 , 191.d-0935 ) , gemini program gn-2016a - q-36 , gn-2015b - q-4 , the uh 2.2 m and the nasa / esa hubble space telescope program 14484 . we thank gemini and hst for directors discretionary time .
we acknowledge eu / fp7-erc grants [ 291222,307260,320360,615929 ] , weizmann - uk making connections grant , stfc ernest rutherford fellowship ( km ) , sofia kovalevskaja award from the alexander von humboldt foundation ( twc ) .
ps1 surveys acknowledge the ps1sc : university of hawaii , mpia heidelberg , mpe garching , johns hopkins university , durham university , university of edinburgh , queen s university belfast , harvard - smithsonian cfa , lcogt , ncu taiwan , stsci , university of maryland , eotvos lorand university , los alamos national laboratory , nsf grant no . ast-1238877 . ,
n. , burgett , w. , chambers , k. , et al .
2010 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol .
7733 , society of photo - optical instrumentation engineers ( spie ) conference series lllllllll name & ra ( j2000 ) & dec ( j2000 ) & disc .
mjd & disc mag & spec mjd & type & spec @xmath55 & classification source and notes + ps15dcq & 03 22 55.83 & + 34 59 23.6 & 57384.29 & 19.99 & 57388.92 & ia & 0.072 & iptf15fgy , @xcite , + & & & & & & & & @xcite + ps15dov@xmath56 & 03 43 57.36 & + 39 17 43.7 & 57384.32 & 19.73 & 57386.32 & ii & 0.016702 & gmos@xmath57 @xcite + ps15dot & 02 11 55.69 & + 13 28 17.8 & 57384.34 & 20.97 & 57386.30 & ii & 0.149 & gmos @xcite + ps15coh &
02 15 58.45 & + 12 14 13.6 & 57384.34 & 17.72 & 57329.22 & ia & 0.020 & old sn , asassn-15rw , + & & & & & & & & iptf15fev @xcite , + & & & & & & & & @xcite + ps15dow & 02 19 42.20 & + 14 09 54.7 & 57384.34 & 20.22 & 57387.21 & ib & 0.05 & gmos @xcite + ps15csf & 02 26 02.24 & + 17 03 40.4 & 57384.35 & 18.68 & 57335.18 & ii & 0.021 & pessto@xmath58 , old sn atel#8264 + ps15dom & 02 34 45.62 & + 18 20 37.7 & 57384.35 & 19.01 & 57390 .
& ii & 0.034 & old sn , psn j02344555 + 182039 , iptf15fdv + & & & & & & & & @xcite + ps15don & 02 37 11.44 & + 19 03 20.2 & 57384.35 & 20.47 & 57388.20 & ia & 0.160 & gmos + ps15doy & 02 47 54.16 & + 21 46 24.0 & 57384.38 & 20.75 & 57388.23 & ia & 0.190 & gmos + ps15dox & 02 40 15.05 & + 22 32 12.1 & 57384.38 & 19.23 & 57389.06 & ia & 0.080 & pessto , @xcite + ps15dpq & 03 09 12.74 & + 27 31 16.9 & 57384.39 & 18.84 & 57389.04 & ia & 0.038 & pessto , iptf15fel , @xcite + & & & & & & & & @xcite + ps15dpa & 02 57 56.02 & + 28 53 37.1 & 57384.40 & 19.51 & 57389.03 & ia & 0.079 & pessto @xcite , + & & & & & & & & master otj025756.02 + 285337 + & & & & & & & & @xcite + ps15dpl & 05 47 45.39 & + 53 36 32.4 & 57384.43 & 19.34 & 57387.40 & ia & 0.03 & sn2016j , asassn-16ah @xcite + ps15dpe & 05 44 42.66 & + 52 24 57.9 & 57384.43 & 19.44 & 57388.25 & ia & 0.057 & gmos + ps16ku & 02 19 06.15 & + 10 37 45.5 & 57385.22 & 20.95 & 57401.24 & ii & 0.061 & snifs + ps15dpn & 02 32 59.75 & + 18 38 07.0 & 57385.23 & 20.69 & 57387.23 & ibn & 0.1747 & gmos , @xcite , + & & & & & & & & iptf15fgl @xcite + & & & & & & & & @xcite , @xcite + ps15doz & 02 53 41.68 & + 27 29 57.8 & 57385.25 & 20.69 & ... & ... & ... & likely sn@xmath59 , slow rise + ps15dpc & 03 55 46.16 & + 38 52 49.6 & 57385.27 & 20.95 & 57387.26 & ii & 0.056 & gmos @xcite + ps15dqc & 05 51 13.43 & + 52 28 18.7 & 57385.29 & 21.16 & ... & ... & ... & likely sn + ps15cvo & 02 20 37.39 & + 17 02 17.9 & 57385.31 & 20.45 & ...
& ... & ... & master022037.36 + 170217.5 , old sn + ps15dpz & 02 40 33.01 & + 23 00 10.8 & 57385.32 & 21.15 & ... & ... & ... & likely sn + ps15dpb & 03 42 23.40 & + 39 14 40.4 & 57385.36 & 20.20 & 57386.43 & ii & 0.041045 & gmos @xcite + ps15dpg & 03 17 18.88 & + 32 20 06.9 & 57385.41 & 20.86 & ... & ... & ... & likely sn + ps15dpx & 06 04 35.54 & + 53 35 25.8 & 57385.56 & 20.76 & 57395.48 & ... & 0.051 & snifs@xmath60 , featureless .
+ ps15dou & 06 03 38.73 & + 54 41 12.1 & 57385.56 & 20.20 & 57395.51 & ii & 0.079 & snifs , @xcite + ps15dpu & 02 40 41.35 & + 16 49 52.0 & 57386.22 & 17.26 & 57396.88 & ii & 0.0292 & asassn-15un , @xcite + ps15dpt & 02 07 34.96 & + 11 03 25.2 & 57386.22 & 20.64 & 57395.36 & ... & ... & snifs , red continuum , possible foreground + ps15dpy & 02 28 22.75 & + 13 59 19.3 & 57386.22 & 21.31 & 57395.39 & ... & ... & snifs , red continuum , possible foreground + ps15dqa & 02 59 41.20 & + 25 14 12.2 & 57386.24 & 20.93 & ... & ... & 0.001251 & likely lbv in ngc1156 + ps16cks & 04 22 33.25 & + 43 36 53.0 & 57386.31 & 21.45 & ... & ... & ... & likely sn + ps15dqd & 05 56 14.60 & + 52 51 55.2 & 57386.37 & 19.92 & ... & ... & ... & likely hostless sn , 0.6@xmath61 fade in 3 days + ps15dqe & 06 05 26.88 & + 54 09 11.3 & 57386.37 & 21.51 & ... & ... & ... & likely sn + ps16kv & 02 22 53.41 & + 19 15 49.9 & 57388.22 & 21.70 & ... & ... & ... & likely sn , host is sdss j022253.48 + 191550.5 + ps16kx & 02 44 42.28 & + 22 36 39.5 & 57389.32 & 21.81 & ... & ... & ... & likely sn + ps16cld & 04 51 13.33 & + 48 59 21.2 & 57392.26 & 21.17 & ... & ... & ... & likely sn + ps16kw & 02 35 50.63 & + 17 33 38.2 & 57394.22 & 21.27 & ... & ... & ... & likely sn + ps16ky & 03 22 34.61 & + 30 36 07.1 & 57397.31 & 20.89 & ... & ... & ... & likely sn + ps16bpe & 02 38 48.30 & + 22 05 56.4 & 57397.34 & 21.56 & ... & ... & ... & likely hostless sn + ps16bpf & 02 56 00.56 & + 24 48 51.8 & 57397.38 & 21.83 & ... & ... & ... & likely sn + ps16bpj & 03 29 06.15 & + 35 39 07.5 & 57397.39 & 21.82 & ... & ... & ... & likely sn + ps16lj & 06 23 09.10 & + 54 38 20.9 & 57397.51 & 20.66 & 57405.24 & ... & 0.088 & snifs , blue continuum , @xmath62 + ps15bpk & 02 37 09.56 & + 22 24 02.4 & 57402.28 & 21.33 & ... & ... & ... & old sn + ps16bpg & 02 56 40.73 & + 27 40 12.0 & 57402.28 & 20.46 & ... & ... & ... & likely sn , rising + ps16bps & 06 04 34.63 & + 53 35 38.8 & 57402.39 & 21.20 & ... & ... & ... & likely sn + ps16bpu@xmath56 & 03 43 57.13 & + 39
17 38.4 & 57402.42 & 19.34 & ... & ... & 0.016702 & likely sn .
offset 20 from position of sn2001i + ps16bpw & 03 06 54.05 & + 28 44 23.2 & 57413.26 & 21.53 & ... & ... & ... & likely sn , young + ps16bqa & 02 38 57.24 & + 18 10 40.4 & 57413.30 & 21.69 & ... & ... & ... & likely sn + ps16aeo & 03 30 46.70 & + 36 38 23.0 & 57414.28 & 19.67 & ... & ... & ... & likely sn + ps16bpz & 06 31 15.13 & + 54 51 52.3 & 57414.35 & 20.18 & ... & ... & ... & likely sn + + ps15dpp & 03 00 39.86 & + 28 15 25.4 & 57384.40 & 20.63 & 57395.42 & ... & ... & sdss j030039.86 + 281525.4 , snifs .
+ ps15dop & 03 17 29.58 & + 29 34 09.2 & 57384.40 & 20.01 & ... & ... & ... & likely agn activity + ps15dpd & 05 09 58.63 & + 50 47 09.4 & 57384.44 & 20.34 & ... & ... & ... & likely stellar + ps15dpo & 02 59 49.56 & + 25 10 30.4 & 57385.25 & 20.55 & ... & ... & ... & agn + ps16li & 06 18 59.16 & + 55 50 55.4 & 57397.51 & 20.18 & ... & ... & ... & likely m - dwarf flare + ps16bpx & 03 50 03.36 & + 37 00 52.1 & 57414.30 & 18.82 & ... & ... & ... & stellar , css100113 - 035003 + 370052 + @xmath57 gmos denotes classification spectra taken for this project with gemini - n and the gmos spectrometer with gratings either r150 or r400 .
+ @xmath58 pessto denotes classification spectra taken for this project with pessto as described in @xcite .
+ @xmath60 snifs denotes classification spectra taken for this project with the snifs instrument on the uh2.2 m telescope as described in @xcite .
+ @xmath63 mjd for gw151226 is 57382.152 + @xmath59 likely sn " means that the transient is not coincident with an observed point source , nor is a known stellar or agn variable , and does have a candidate host galaxy nearby and a lightcurve that is consistent with being a normal sn .
+ @xmath56 ps15dov and ps16bpu exploded in the same galaxy ugc2836 . which also hosted sn2001i and sn2003ih . | we present a search for an electromagnetic counterpart of the gravitational wave source gw151226 . using the pan - starrs1 telescope we mapped out 290 square degrees in the optical @xmath0 filter starting 11.5hr after the ligo information release and lasting for a further 28 days .
the first observations started 49.5hr after the time of the gw151226 detection .
we typically reached sensitivity limits of @xmath0@xmath1 and covered 26.5% of the ligo probability skymap .
we supplemented this with atlas survey data , reaching 31% of the probabilty region to shallower depths of @xmath2 .
we found 49 extragalactic transients ( that are not obviously agn ) , including a faint transient in a galaxy at 7mpc ( a luminous blue variable outburst ) plus a rapidly decaying m - dwarf flare .
spectral classification of 20 other transient events showed them all to be supernovae .
we found an unusual transient , ps15dpn , with an explosion date temporally coincident with gw151226 which evolved into a type ibn supernova .
the redshift of the transient is secure at @xmath3 and we find it unlikely to be linked , since the luminosity distance has a negligible probability of being consistent with that of gw151226 . in the 290 square degrees
surveyed we therefore do not find a likely counterpart .
however we show that our survey strategy would be sensitive to ns - ns mergers producing kilonovae at @xmath4mpc , which is promising for future ligo / virgo searches . |
both the van der waals and casimir forces are quantum phenomena originating from the interaction of electromagnetic fluctuations with matter .
one difference between them is that the van der waals force is nonrelativistic , whereas the casimir force takes full account of the relativistic retardation effects .
thus , in fact there exists just one force of a fluctuation nature . at shortest sepatations up to a few nanometers between the interacting bodies , where retardation is negligibly small , it is referred to as the _ van der waals _ force . at larger separations
it is called the _ casimir _
force @xcite .
the casimir force becomes significantly large if measured at separations below one micrometer . at separations below 100
nm it competes with characteristic electric forces acting between the constituting elements of microelectromechanical and nanoelectromechanical devices .
the attractive casimir force leads to pull - in and stiction @xcite which may result in the collapse of a microdevice .
that is why the casimir force has attracted much recent attention in nanotechnology . on the one hand
, it can be used as an actuating force @xcite in place of electric forces . on the other hand
, it plays the role of some kind friction in micrometer scales which can not be diminished by means of lubrication . in this respect ,
the possibility of repulsive casimir force attracts much public attention as the mean to cancel this friction .
it is commonly known that the unified description of both the van der waals and casimir forces is given by the lifshitz theory @xcite . during more than fourty years after
this theory was created it was lacking detailed experimental confirmation due to the absence of sufficiently precise measurements . during the last decade ,
however , many experiments measuring the casimir force have been performed ( see the recent review @xcite ) .
the comparison between the measurement data of the most precise experiments and computational results on the basis of the lifshitz theory has led to a puzzle .
it turned out that the lifshitz theory is in agreement with the data only under a condition that for metallic test bodies the relaxation properties of conduction electrons are not accounted for , whereas for dielectrics charge carriers are completely discarded ( controversial discussions concerning these results are reflected in ref .
modern overview of both experimental and theoretical aspects of the casimir force can be found in the monograph @xcite .
the present paper reviews several recent experimental and theoretical results in the casimir effect which are of much importance for understanding of delicate physical mechanisms of fluctuation - induced forces and for their technological application in new generations of micro- and nanoelectromechanical systems . in sec . 2 we briefly discuss the standard formalism of the lifshitz theory , its generalization for the case of nonplanar geometries , and modifications aimed to resolve the above - mentioned puzzle .
section 3 is devoted to the comparison of theoretical results with the experimental data of three landmark experiments .
these experiments were performed with metallic @xcite , semiconductor @xcite , and dielectric @xcite test bodies . in sec .
4 the possibilities to control the magnitude and sign of the casimir force by using materials with different electric and magnetic properties are considered .
these include the variation of the density of free charge carriers with laser light , the use of three - layer systems and phase transitions , investigation of the casimir force between ferromagnetic dielectrics etc .
special attention is devoted in sec . 5 to the lateral casimir force arising between sinusoidally corrugated surfaces , as predicted theoretically in ref .
@xcite and experimentally demonstrated in refs .
it was shown quite recently @xcite that this phenomenon has potential for applications in nanotechnology .
our conclusions and discussion are contained in sec .
the lifshitz theory @xcite expresses the free energy and pressure of the van der waals and casimir interaction in the configuration of two parallel semispaces ( thick plates ) separated with a gap of width @xmath0 at temperature @xmath1 in the following form :
@xmath2 here , @xmath3 is the boltzmann constant , @xmath4 with @xmath5 are the matsubara frequencies , and the terms with @xmath6 in the primed sums are multiplied by 1/2 .
the functions @xmath7 and @xmath8 are expressed in terms of reflection coefficients on the two semispaces @xmath9 and @xmath10 for two independent polarizations of the electromagnetic field , transverse magnetic ( @xmath11 ) and transverse electric ( @xmath12 ) , @xmath13 , \nonumber \\ & & \phi_p(x)=\int_{0}^{\infty}k_{\bot}dk_{\bot}q\sum_{\alpha } \left[\frac{e^{2aq}}{r_{\alpha}^{(1)}(ix , k_{\bot } ) r_{\alpha}^{(2)}(ix , k_{\bot})}-1\right]^{-1 } , \label{eq2}\end{aligned}\ ] ] where @xmath14 and @xmath15 is the projection of a wave vector on the plane of plates .
the reflection coefficients on homogeneous media described by the frequency - dependent dielectric permittivity @xmath16 and magnetic permeability @xmath17 ( @xmath18 ) are given by @xmath19 where @xmath20 the lifshitz - type formulas similar to ( [ eq1 ] ) hold also for the casimir - polder free energy and force acting between an atom with dynamic electric polarizability @xmath21 and magnetic susceptibility @xmath22 and a wall decribed by @xmath23 and @xmath24 @xmath25 where @xmath26 .
the function @xmath27 is defined as @xcite @xmath28\right . \label{eq6 } \\ & -&\left .
\frac{x^2}{q^2c^2 } \left[\alpha(ix)+\beta(ix)\right]\,\left[r_{\rm tm}(ix , k_{\bot})+ r_{\rm te}(ix , k_{\bot})\right]\right\}. \nonumber\end{aligned}\ ] ] due to the presence of only one wall , the upper index on the reflection coefficients is omitted .
using the so - called _ proximity force approximation _
( pfa ) , the casimir force acting between a sphere and a plate can be expressed as @xmath29 where the free energy per unit area for two parallel semispaces is defined in eq .
( [ eq1 ] ) . under the condition that @xmath30 , where @xmath31 is the sphere radius and @xmath0 is the closest separation between the surfaces of a sphere and a plate , the relative error of the approximate eq .
( [ eq7 ] ) is less than @xmath32 @xcite .
equation ( [ eq7 ] ) is widely used for the comparison of the experimental data obtained from the measurements of the casimir force between a sphere and a plate with the predictions of the lifshitz theory
. the original formulation of the lifshitz theory for two parallel semispaces separated with a gap can be simply generalized for an arbitrary number of planar layers of different thickness made of arbitrary materials @xcite . however , the obtaining of exact formulas for the casimir free energy and force between bodies with nonplanar surfaces was challenging during the last few decades .
the problem was solved in 2006 when several equivalent lifshitz - type formulas were obtained @xcite using the scattering theory and the formalism of functional determinants .
specifically , in one of the developed formalisms the casimir free energy for two separated bodies of arbitrary shape labeled by the indices ( 1 ) and ( 2 ) can be presented in the form @xcite @xmath33 . \label{eq8}\ ] ] here , @xmath34 is the operator for the free space green s function with the matrix elements @xmath35 , where @xmath36 belongs to the body ( 1 ) and @xmath37 belongs to the body ( 2 ) . @xmath38
[ @xmath39 is the operator of @xmath1 matrix for the body ( 1 ) [ body ( 2 ) ] .
using such representations , it was shown that for an ideal metal sphere above an ideal metal plate at room temperature the exact results for the casimir force and for the thermal correction to it coincide with respective results obtained using the pfa in the zeroth order of the small parameter @xmath32 @xcite .
calculation of the casimir force between real material bodies using the lifshitz theory requires the availability of the respective @xmath23 and @xmath24 .
we reserve the discussion of @xmath24 for sec . 4 and briefly consider here the dielectric permittivity for dielectrics and metals .
it is common knowledge that all condensed bodies are either dielectrics or metals depending on the behavior of their dc conductivity @xmath40 when temperature vanishes @xcite . for dielectrics @xmath41
whereas for metals @xmath42 . in this regard semiconductors are also either dielectrics or metals ( specifically , intrinsic semiconductors are dielectrics , whereas doped semiconductors with dopant concentration above the critical value are metals ) . in the application of the lifshitz theory to dielectric materials the small dc conductivity arising at @xmath43
is usually neglected . in so doing the dielectric permittivities of dielectric plates
are determined by the core electrons @xcite @xmath44 where @xmath45 are the oscillator frequencies , @xmath46 are the oscillator strengths , @xmath47 are the damping parameters , and @xmath48 is the number of oscillators .
note that the parameters of oscillators in eq .
( [ eq9 ] ) can be temperature - dependent .
thus , in addition to the explicit dependence on the temperature in eqs .
( [ eq1 ] ) and ( [ eq5 ] ) through the multiple @xmath1 and through the matsubara frequencies , there is an implicit dependence through the dielectric permittivity .
for si such an implicit dependence was investigated in ref .
@xcite .
however , if one takes into account the existence of free charge carriers at @xmath43 , the dielectric permittivity of dielectric plates can be represented as @xcite @xmath49 it has been shown @xcite that the casimir entropy calculated using the free energy ( [ eq1 ] ) combined with the dielectric permittivity ( [ eq9 ] ) satisfies an important thermodynamic constraint @xmath50 when @xmath51 ( the nernst heat theorem ) . if , however , the dc conductivity of dielectric material is taken into account [ i.e. , the dielectric permittivity ( [ eq10 ] ) is used ] we get @xcite @xmath52>0 , \label{eq12}\ ] ] where @xmath53 @xmath54 is the riemann zeta function , and @xmath55 is the polylogarithm function . keeping in mind that real dielectric materials possess dc conductivity , one arrives to the conclusion that for two parallel dielectric semispaces the lifshitz theory violates the nernst teorem .
a similar difficulty arises in the application of the lifshitz theory to real metals .
it is customary to describe the dielectric properties of metallic plates by means of the drude model @xmath56}\ , , \label{eq13}\ ] ] where @xmath57 are the plasma frequencies and @xmath58 are the relaxation parameters .
this model is well applicable in the wide frequency region from quasistatic frequencies ( the normal skin effect ) to infrared optics .
it was shown @xcite that for perfect crystal lattices [ where @xmath59 as @xmath60 the casimir entropy calculated using the free energy ( [ eq1 ] ) combined with the drude model ( [ eq13 ] ) , violates the nernst heat theorem @xmath61<0 . \label{eq14a}\ ] ] notice that for drude metals with impurities there is a nonzero residual relaxation @xmath62 . in this case
the casimir entropy jumps abruptly to zero at @xmath63k starting from a negative value ( [ eq14a ] ) , so that the nernst theorem is formully satisfied @xcite .
this , however , does not solve the problem .
the point is that a perfect crystal lattice is a truly equilibrium system with a nondegenerate dynamical state of the lowest energy .
thus , according to quantum statistical physics , the casimir entropy calculated for a perfect crystal lattice must be equal to zero @xcite .
it is quite another matter when the nondissipative plasma model @xmath64 is used instead of ( [ eq13 ] ) to describe metallic plates . in this case
the nernst heat theorem is satisfied @xcite , i.e. @xmath65 when @xmath51 .
it was proposed @xcite that the inclusion of relaxation properties of free charge carriers in the model of dielectric response violates thermal equilibrium which is the basic applicability condition of the lifshitz theory .
this is because the drift electric current leads to a violation of time - reversal symmetry and requires the introduction of the unidirectional flux of heat from the casimir plates to the heat reservoir to preserve the temperature constant .
consensus on the problem of why the lifshitz theory is thermodynamically inconsistent in the presence of free charge carriers is not yet achieved . to complete a brief survey of the lifshitz theory , we mention recent attempts to modify the tm reflection coefficients by taking into account the effect of screening @xcite .
this approach includes both the drift and diffusion currents of free charge carriers .
eventually the modified tm reflection coefficient is obtained through the use of boltzmann transport equation and takes the form @xcite @xmath66 here , the dielectric permittivities of the plates are given by @xmath67 where permittivities of the bound core electrons @xmath68 are defined in eq .
( [ eq9 ] ) .
the quantity @xmath69 is given by @xmath70^{1/2 } , \label{eq18}\ ] ] where @xmath71 are the screening lengths of plate materials @xcite defined differently for dielectrics and metals ( the debye - hckel and tomas - fermi screening lengths when the charge carriers are described by the maxwell - boltzmann and fermi - dirac statistics , respectively ) .
the te reflection coefficient in the proposed approach remains unchanged .
it was shown @xcite , however , that the modification of the tm reflection coefficient in accordance with eq .
( [ eq16 ] ) also leads to the violation of the nernst theorem for all dielectric materials whose charge carrier density does not vanish when @xmath51 and @xmath40 vanishes due to the vanishing mobility ( for instance , for glasses with ionic conductivity @xcite or doped semiconductors with dopant concentration below the critical value @xcite ) . for metals with perfect crystal lattices
the lifshitz theory with the modified tm reflection coefficient violates the nernst theorem in the same way as when the standard drude model approach is used @xcite .
these conclusions were disputed in refs .
@xcite , but the argumentation used was shown to be inadequate in refs .
therefore , the problem of consistency of the lifshitz theory with thermodynamics for real materials remains unsolved .
there are about 25 experiments on measuring the casimir force in the modern stage of the work on the subject ( see the review @xcite for more details ) . in almost all experiments
at least an attempt of comparing the measurement data with theory was undertaken . in this section ,
we consider only three experiments each of which , when carefully compared with the lifshitz theory , leads to unexpected conclusions .
the first of this landmark experiments was performed by r. s. decca and his co - workers .
it was repeated three times , in 2003 @xcite , 2005 @xcite , and in 2007 @xcite with major improvements in sample preparation and in the experimental precision in each repetition .
the discussion below is related to the most precise experiment of 2007 @xcite , where the gradient of the casimir force between an au - coated sphere of @xmath72 m radius and a plate of micromachined oscillator was measured at @xmath73k in the dynamic regime ( see fig . 1 for the schematic diagram of a setup ) .
in fact it was the shift of the resonant frequency of a micromachined oscillator that was measured .
this shift is proportional to the gradient of the casimir force between a sphere and a plate . considering the negative derivative with respect to @xmath0 of both sides of eq .
( [ eq7 ] ) , one obtains @xmath74 thus , the use of the pfa allows interpretation of the experiment by decca et al . as an indirect measurement of the casimir pressure between two parallel plates . in this experiment
all errors were determined at a 95% confidence level , and the random errors were made much less than the systematic errors , so that only the latter determined the total experimental error . as a result , the total relative experimental error of the casimir pressure varied from 0.19% at @xmath75 nm to 0.9% at @xmath76 nm and to 9.0% at the largest separation @xmath77 nm .
this includes the error of the pfa which was used to recalculate the experimental data for the gradient of the casimir force into the casimir pressure . in the most conservative way the error of the pfa was taken to be equal to @xmath32 ( specially performed experiment @xcite and theoretical computations @xcite for a sphere and a plate made of real metals described by the plasma and drude models
show that this error is in fact smaller than @xmath32 ) .
computations of the casimir pressure between au plates was performed by eq .
( [ eq1 ] ) with @xmath78 using the tabulated optical data @xcite for the complex index of refraction @xmath79 . for this purpose the imaginary part of the dielectric permittivity of au was found as @xmath80 at frequencies where the optical data are available and extrapolated to lower frequencies by means of the drude model ( [ eq13 ] ) with the parameters @xmath81ev , @xmath82ev @xcite .
then @xmath83 for au was obtained using the kramers - kronig relation @xmath84 this calculational approach corresponds to the use of the drude model , as described in sec . 2 , but takes into account the role of core electrons .
keeping in mind that problems discussed in sec .
2 originate from the zero - frequency term of the lifshitz formula , one might expect that related difficulties should appear here as well .
for the comparisom with the experimental data , the surface roughness on both the sphere and the plate was studied with the help of an atomic force microscope and taken into account in a nonmultiplicative way by means of geometrical averaging @xcite . another approach to calculate the casimir pressure between real metals with account of core
electrons uses the generalized plasma - like model for the permittivity @xmath85 where @xmath86 is defined in eq .
( [ eq9 ] ) and the parameters of oscillators for au with @xmath87 were determined in ref .
the surface roughness is taken into account as described above .
now we present a typical comparison between the experimental data and theory . in fig .
2(a , b ) the mean casimir pressure together with the absolute experimental errors in the separation and pressure measured at different separations is shown as crosses . in fig .
2(a ) the theoretical casimir pressure computed using the generalized plasma - like model and the optical data of ref .
@xcite extrapolated by the drude model is shown by the light - gray and dark - gray bands , respectively .
the width of the bands indicates the total theoretical error . as can be seen in fig .
2(a ) , the plasma - like model is consistent with the data , whereas a theory using the extrapolation to low frequencies by means of the drude model is excluded by the data at a 95% confidence level .
as was noted in the literature @xcite , au films prepared using different technology may possess different sets of optical data leading to the values of @xmath88 varying from 6.85 to 9.0ev .
2(b ) the theoretical casimir pressure as a function of separation computed with all sets of optical data available is shown as the dark - gray band .
it is seen that the use of any alternative value of @xmath88 makes the disagreement between the drude model approach and the data even more evident .
one more recent experiment @xcite incorporated the determination of the casimir pressure by means of a micromachined oscillator with simultaneous measurement of the optical data of au films .
it was shown that while the casimir pressures coincide with those measured previously in refs .
@xcite , the optical data closely follow those tabulated in ref .
@xcite .
another method for comparing theory with experiment considers the differences between theoretical and mean experimental casimir pressures , @xmath89 , and the 95% confidence interval @xmath90 $ ] to which 95% of these differences must belong if the theory is consistent with the data . in fig .
3(a ) , the differences of the theoretically computed pressures using the generalized plasma - like model and mean experimental casimir pressures are shown as dots .
the solid lines are formed by the ends of the confidence intervals .
as can be seen in fig .
3(a ) , the lifshitz theory using the generalized plasma - like model is consistent with the data .
a similar comparison is made with the theoretical casimir pressures computed using the tabulated optical data extrapolated by the drude model . in fig .
3(b ) the differences @xmath89 are indicated as two sets of dots computed using @xmath91ev ( the upper set ) and @xmath81ev ( the lower set ) .
the pressure differences for all other values of the plasma frequency of au considered in ref .
@xcite are between the two sets of dots in fig .
thus , theory using the drude model is experimentally excluded over the entire measurement range from 162 to 746 nm at a 95% confidence level .
note that within a more narrow range of separations from 210 to 620 nm the theoretical approach using the drude model is excluded by the data at a 99.9% confidence level @xcite .
the same data exclude @xcite the lifshitz theory at zero temperature combined with the drude model .
this exclusion is made at a 70% confidence level with the optical data of ref .
@xcite and at 95% confidence level with the alternative optical data . the experimental data of ref .
@xcite have been also compared with theoretical predictions obtained using the modified tm reflection coefficient ( [ eq16 ] ) . in so doing it
was supposed that electrons in au are described by fermi - dirac statistics .
almost the same computational results for the casimir pressure were obtained @xcite , as by using the drude model .
thus , the lifshitz theory with the modified tm reflection coefficient is excluded by the experiment of refs .
@xcite .
the second landmark experiment was performed by u. mohideen and his co - workers @xcite . in this experiment ,
an atomic force microscope was employed to measure the change in the casimir force between an au - coated sphere ( @xmath92 m ) and a si membrane at @xmath73k in the presence and in the absence of incident laser light on a membrane ( see schematic of the experimental setup in fig .
4 ) .
the experiment was repeated three times with three different absorbed laser powers .
the values of the maximum and minimum absorbed powers were equal to @xmath93mw and @xmath94mw . in the absence of laser light ,
the charge carrier density of high - resistivity si was @xmath95 . in the presence of light
, the charge carrier densities related to the maximum and minimum absorbed powers took the values @xmath96 and @xmath97 ( the errors are indicated at a 95% confidence level @xcite ) .
the difference of the theoretical casimir forces in the presence and in the absence of laser light , @xmath98 was calculated using eq .
( [ eq1 ] ) with @xmath78 and eq .
( [ eq7 ] ) , taking into account the surface roughness by means of geometrical averaging .
the dielectric permittivity of the au - coated sphere @xmath99 is described by eq .
( [ eq21 ] ) .
the dielectric permittivity of the si plate @xmath100 depends on the concentration of charge carriers . in the absence of charge carriers ( infinitely high resistivity )
it was computed by using the tabulated optical data @xcite and the kramers - kronig relation .
the obtained results , @xmath101 , are shown by the dashed line in fig .
the dielectric permittivity of high - resistivity plate used in the experiment in the absence of laser light , @xmath102 , can be presented as the sum of @xmath101 and the second term on the right - hand side of eq .
( [ eq10 ] ) taking the dc conductivity into account . in fig .
5(a ) the quantity @xmath102 is shown by the dotted line .
similarly , the dielectric permittivity of low - resistivity si with different absorbed powers , @xmath103 , are shown by the solid lines a and b taking into account both electrons and holes @xcite .
k. ( b ) the experimental differences in the casimir force with their experimental errors are shown as crosses ( the absorbed power is equal to 4.7mw ) .
the solid and dotted lines represent the theoretical differences computed at @xmath73k using the model with a finite static permittivity of high - resistivity si , but different models for si in the presence of light ( see text for further discussion ) .
the dashed line represents the theoretical differences computed including the dc conductivity of high - resistivity si.,width=907 ] now we are in a position to compare the experimental data with the predictions of the lifshitz theory . in fig .
5(b ) the mean experimental differences , @xmath104 , of the casimir forces in the presence and in the absence of laser light for the absorbed power @xmath105 are shown as crosses .
the arms of the crosses represent total experimental errors in the measurement of the difference force and separation distance , determined at a 95% confidence level .
the theoretical differences of the casimir force ( [ eq22 ] ) were first computed using the dielectric permittivity @xmath102 in the absence of light which takes the dc conductivity of a si plate into account . in the presence of light the permittivity
@xmath106 has been used .
the computational results are presented by the dashed line in fig .
as can be seen in the figure , the prediction of the lifshitz theory taking into account the dc conductivity of high - resistivity si is experimentally excluded .
then the theoretical differences ( [ eq22 ] ) were computed using the dielectric permittivity @xmath101 in the dark phase which disregards the dc conductivity of si . in the presence of light
the permittivity @xmath106 was again used . in this case
the computational results are shown by the solid line in fig .
5(b ) which is consistent with the data .
note that the dielectric permittivity @xmath106 used in the presence of light describes charge carriers in the low - resistivity si by means of the drude model . if to take into account that @xmath107 is larger than the critical value of charge carrier density indicating the dielectric to metal phase transition , it would be reasonable to use the description by means of the generalized plasma - like model in the presence of light ( see sec . 2 ) .
the computational results obtained within this approach @xcite are presented by the dotted line in fig .
it is seen that the dotted line is consistent with the data , as well as the solid line .
thus , this experiment is not of sufficient precision to discriminate between the description of charge carriers in metallic phase by means of the drude and plasma models ( this was done by the experiment of decca et al . considered in sec .
the main new and unexpected result of the experiment by mohideen et al .
is that the inclusion of the dc conductivity in a dielectric phase ( high - resistivity si ) results in the contradiction of the lifshitz theory with the data . because of this
, phenomenologically one may conclude @xcite that in applications of the lifshitz theory to dielectric materials the dc conductivity should be disregarded .
note that the contribution of dc conductivity to the dielectric permittivity of dielectrics was also disregarded when studying nonequilibrium thermal fluctuations of the electric field in the presence of a dc current @xcite . .
the two theoretical bands lie between the solid and dashed lines ( see text for details ) .
( b ) theoretical minus mean experimental differences of the casimir force versus separation are shown as dots .
for the dots labeled 1 , the theoretical results are computed with a finite static dielectric permittivity of high resistivity si in the absence of laser pulse and using the plasma model to describe the charge carriers when the laser pulse is on . for dots labeled 2 ,
the theoretical results are computed in the framework of the modified lifshitz theory @xcite .
the solid lines indicate 70% confidence intervals including all experimental and theoretical errors.,width=718 ] the experimental data of the optical modulation experiment was compared with the modified lifshitz theory using the tm reflection coefficient ( [ eq16 ] ) proposed in refs .
it was found that predictions of the modified lifshitz theory for the si plate in a dielectric phase can not be excluded at a 95% confidence level .
because of this , the comparison has been made at a 70% confidence level @xcite . in fig .
6(a ) the experimental data for the measured difference casimir force in the presence and in the absence of light are shown as crosses . when compared to fig .
5(b ) , the arms of the crosses are made relatively shorter , as required by a lower confidence level .
the prediction of the modified lifshitz theory for the absorbed power @xmath105 is shown as a band bounded by the two dashed lines .
the width of this band is found from the theoretical error determined mostly by the error in @xmath107 ( at a 70% confidence level this error is equal to @xmath108 to compare with @xmath109 at a 95% confidence level ) .
the prediction of the standard lifshitz theory with dc conductivity of si in the absence of light disregarded is shown by the band bounded by two solid lines .
the width of this band also corresponds to an error in @xmath107 determined at a 70% confidence level . as can be seen in fig .
6(a ) , the predictions of the modified lifshitz theory @xcite are excluded by the data at a 70% confidence level , whereas the standard lifshitz theory with disregarded dc conductivity of dielectric si is consistent with the data .
the same conclusion can be obtained when using another approach to the comparison between experiment and theory similar to that in fig .
3 . in fig .
6(b ) the differences between theoretical and mean experimental difference casimir forces , @xmath110 , at different separations are plotted as the two sets of dots .
the set labeled 1 is computed using the standard lifshitz theory with the dc conductivity of dielectric si disregarded .
the set labeled 2 is computed using the modified lifshitz theory .
the two solid lines indicate the borders of a 70% confidence intervals determined with account of all experimental and theoretical errors .
it is seen that the standard lifshitz theory with the dc conductivity disregarded is consistent with the data , whereas the modified lifshitz theory @xcite is excluded by the measurement results of the optical modulation experiment at a 70% confidence level .
the third landmark experiment is on measuring the casimir - polder force between ground state @xmath111rb atoms and a fused silica plate .
it was performed by cornell and his co - workers @xcite .
atoms of @xmath111rb belonged to the bose - einstein condensate separated by a distance of a few micrometers from a wall .
the experiment was of dynamic - type .
the dipole oscillations in the direction perpendicular to the plate with the frequency @xmath112 were excited in a condensate , and the relative shift of this frequency @xmath113 due to the casimir - polder interaction was measured precisely .
this shift can be recalculated @xcite into the casimir - polder force ( [ eq5 ] ) with @xmath78 , @xmath114 leading to its indirect measurement .
this experiment was repeated three times , once in thermal equilibrium , when the temperature of the plate was equal to the environment temperature , @xmath115k , and two times with @xmath116 ( in the latter case there are additions to the casimir - polder force ( [ eq5 ] ) calculated in ref .
the relative frequency shift ( [ eq23 ] ) was measured at several atom - plate separations in the region from 7 to @xmath117 m .
keeping in mind that the comparison of the frequency shift with the casimir - polder force includes an averaging over the condensate cloud , the role of atom - plate separation is in fact played by the distance between the plate and the center of mass of the condensate .
k and out of thermal equilibrium ( b ) with @xmath118k and @xmath119k and ( c ) with @xmath120k and @xmath119k .
computations are done by neglecting ( solid lines ) and including ( dashed lines ) the dc conductivity of the dielectric plate .
the experimental data are shown as crosses .
, width=831 ] in fig .
7(a ) the experimental data for the relative frequency shift @xmath121 in thermal equilibrium at separations below @xmath122 m are shown as crosses @xcite .
the arms of the crosses indicate total experimental errors in the measurement of @xmath121 and separation distances at a 66% confidence level found at each experimental point separately using the analysis of ref .
@xcite . to compare the experimental data with theory , in ref .
@xcite the relative frequency shift ( [ eq23 ] ) was computed using the lifshitz formula ( [ eq5 ] ) for the casimir - polder force and an averaging over the condensate cloud . in so doing the static value for the polarizability of rb atom @xmath123 can be used .
the dielectric permittivity of fused silica @xmath124 , as a function of @xmath125 , can be calculated @xcite using the tabulated optical data @xcite and the kramers - kronig relation ( [ eq16 ] ) . in the original ref . @xcite it was assumed that the fused silica is a true dielectric of infinitely high resistivity .
thus , the dc conductivity of the plate was disregarded . under this assumption
the obtained theoretical results @xcite are shown by the solid line in fig .
it is seen that the lifshitz theory of the casimir - polder interaction with the dc conductivity of fused silica disregarded is consistent with the measurement data .
it should be noted , however , that although fused silica is a good insulator , it possesses a nonzero conductivity at @xmath126k .
this conductivity is ionic in nature and is determined by the concentration of impurities ( alkali ions ) . in ref . @xcite the computations of ref .
@xcite were repeated both disregarding and including the dc conductivity of fused silica .
when the dc conductivity was disregarded , the results of ref .
@xcite given by the solid line were reproduced .
when , however , the dc conductivity was included , quite different results were obtained shown by the dashed line in fig .
it is seen that with inclusion of the dc conductivity the lifshitz theory is inconsistent with at least two experimental points at shortest separations .
the second and third repetitions of the experiment of ref .
@xcite were performed with the same temperature of environment , @xmath119k , but with increased temperatures of the plate , @xmath118k and 605k , respectively .
the obtained experimental data are shown as crosses in fig .
7(b ) for @xmath118k and fig .
7(c ) for @xmath120k .
the computational results of ref .
@xcite disregarding the dc conductivity of fused silica are shown as the solid lines in fig . 7(b , c ) .
they were reproduced in ref .
@xcite under an assumption that the dc conductivity of fused silica is omitted . as is seen in fig .
7(b , c ) , the solid lines are in a good agreement with the experimental data . in ref .
@xcite computations of the relative frequency shift @xmath121 for the second and third repetitions of the experiment with increased temperature of the plate were performed taking the dc conductivity of fused silica into account .
the computational results shown by the dashed lines in fig .
7(b , c ) differ dramatically from the results obtained with the dc conductivity disregarded . as can be seen in fig .
7(b ) , the three experimental points for @xmath118k exclude the dashed line and the other two only touch it . as to the dashed line in fig .
7(c ) , it is seen that all data points at @xmath120k exclude the theoretical prediction computed taking the dc conductivity of fused silica into account .
thus , measurements of the thermal casimir force between @xmath127rb atoms and fused silica plate suggest that in the application of the lifshitz theory to dielectric materials the dc conductivity of these materials should be omitted .
in fact , the results of all three landmark experiments performed by three different experimental groups are in agreement . according to these results ,
the relaxation properties of free charge carriers are irrelevant to the phenomenon of dispersion forces and should be disregarded . for metals characterized by a high density of charge carriers , they should be accounted for by means of the plasma model . for
dielectrics charge carriers should be ignored .
if this phenomenological prescription @xcite is used , the lifshitz theory agrees well with all available experimental data .
applications of the casimir force to nanotechnology require some operational means to modify the force magnitude at a fixed separation distance and even change the force sign with varying separation .
this would permit to realize the actuation of devices with the casimir force at a nanoscale and , when necessary , avoid such undesired effects as pull - in and stiction mentioned in sec . 1 .
below we briefly discuss different possibilities to control the casimir force proposed in the literature some of which were already tested experimentally .
special attention is paid to magnetic materials which could significantly widen the role of casimir forces in applied science . here
, we do not discuss the casimir force between metamaterials because the application of the lifshitz formula to such materials is questionable , and corresponding experimental data are not yet available . in spite of repeated speculations on the possibility of casimir repulsion ,
it was shown @xcite that the casimir force between two plates made of metamaterials and separated by a vacuum gap is always attractive . to modify the magnitude of the casimir force
, the dielectric properties of interacting bodies should be changed over a wide frequency region .
the most apparent possibility of decreasing the magnitude of the casimir force , as compared with the case of two metallic test bodies , is to make one of them of a semiconductor or dielectric ( or at least to cover them with sufficiently thick dielectric coating ) . as an example , in the experiment of refs .
@xcite the casimir force acting between a au coated sphere and si plate at @xmath128 nm in vacuum was of about 30% less than between the same au sphere and a au plate . in the experiment of ref .
@xcite the casimir forces acting between a au sphere and two si plates with significantly different dopant concentrations were sequentially measured .
the density of charge carriers in the first plate was @xmath129 and in the second plate @xmath130 .
it was found that by choosing plates with such different concentrations of charge carriers the casimir force acting between a au sphere and a si plate can vary by 5% and 8% at separations @xmath131 and 100 nm , respectively . in ref .
@xcite the measured values of the casimir force between au sphere and au plate were compared with respective measurements resulted from a au sphere and a glass plate covered with a 190 nm layer of indium tin oxide which is a conductor at low frequencies ( with respect to the casimir force this material was first discussed in ref .
@xcite ) . within the separation region from 50 to 150 nm the casimir force between a sphere and a plate covered with indium tin oxide was shown to be 40%50% less than between the same sphere and a au plate .
the use of test bodies made of dielectric material or coated with dielectric layers allows to decrease the magnitude of the casimir force but does not allow to change this magnitude periodically .
the latter was realized in the optical modulation experiment of refs .
@xcite considered in sec . 3.2 . in this experiment ,
the differences of the casimir forces in the presence and in the absence of laser light on a si plate are equal to 3% and 6.5% of the casimir force in the absence of light at separations @xmath132 and 150 nm , respectively .
it is important that here the magnitude of the casimir force varies periodically as long as the laser operated in the pulse mode is switched on .
optical modulation of the casimir force in vacuum makes it possible to periodically change the magnitude of the casimir force but does not allow to replace attraction with repulsion .
considerable opportunities for nanotechnology would be opened up if pulsating casimir plates were possible , moving back and forth under the action of the casimir force without use of micromechanical springs .
this can be achieved only by using both attractive and repulsive casimir forces . in this respect
, it should be noted that there is the case where the casimir repulsion is well understood , i.e. , the repulsion between the two plates described by the dielectric permittivities @xmath133 separated with a liquid layer having the dielectric permittivity @xmath134 such that @xcite @xmath135 the casimir repulsion in three - layer systems was observed in refs .
@xcite ( see also refs .
@xcite concerning additional difficulties arising for measurements of the casimir force in liquids ) .
recently it was shown @xcite that the illumination of one ( si ) plate in three - layer systems au ethanol
si , si ethanol si , and @xmath136-al@xmath137o@xmath138ethanol si with laser pulses can change the casimir attraction to the casimir repulsion and vice versa .
this allows to combine the advantages of the optical modulation with the casimir repulsion in three - layer systems . as a result
, one can obtain attraction and repulsion interchangeably at regular intervals .
calculations show @xcite that in the system au ethanol
si the casimir force is repulsive at @xmath139 nm .
the illumination of the si plate with laser light changes this repulsion for attraction . in the system si ethanol
si , the casimir force between the si plates is attractive in the absence of laser light .
however , when one of the two si plates is illuminated , the attraction is replaced with repulsion at separations @xmath140 nm @xcite . in both these systems
the magnitude of the repulsive pressure is several times less than the magnitude of the attractive pressure at the same separation .
however , in the three - layer system @xmath136-al@xmath137o@xmath138ethanol
si , where the si plate is illuminated with laser pulses , the light - induced casimir repulsion is of the same order of magnitude as the casimir attraction .
specifically , in this system in the presence of light the casimir repulsion arises at @xmath141 nm .
thus , at @xmath142 nm the attractive casimir pressure in the absence of light is equal to about 75mpa , whereas the repulsive pressure in the presence of light is equal to 50mpa .
the dielectric properties undergo changes during different phase transitions .
it was shown @xcite that the change in the casimir free energy associated with the phase transition of a metal to the superconducting state is very small .
nevertheless , according to refs .
@xcite , the magnitude of this change is comparable to the condensation energy of a superconducting film and causes a measurable increase in the value of the critical magnetic field .
it was also proposed @xcite to measure the change of the casimir force in a superconducting cavity due to a small change of temperature .
considerable change of the dielectric properties occurs in semiconductor materials that undergo a dielectric - metal phase transition with an increase in temperature .
it was proposed @xcite to measure the change of the casimir force acting between an au - coated sphere and a vanadium dioxide ( vo@xmath137 ) film deposited on a sapphire substrate , where the vo@xmath137 undergoes a phase transition . at @xmath143c thin vo@xmath137 films
undergo an abrupt transition from a dielectric - type semiconductor to a metallic phase . during this phase transition
the resistivity of the film decreases by a factor @xmath144 from 10 to @xmath145 cm .
the increase in temperature necessary for the phase transition can be induced by laser light .
thus , a setup similar to the one described in sec .
3.2 can be used .
preliminary theoretical results using the optical data for vo@xmath137 films show @xcite that the difference casimir force between a au sphere and vo@xmath137 film after and before the phase transition is sensitive to the model of conductivity properties used in computations .
thus , this experiment may help in the resolution of the problems discussed in secs . 2 and 3 .
another type of dielectric - metal phase transitions leading to the change of dielectric properties is a transition from amorphous to crystalline state in aginsbte . in ref .
@xcite the gradient of the casimir force between a au coated sphere of @xmath146 m diameter and @xmath147 m thick aginsbte films deposited onto al - coated si substrates was measured using an atomic force microscope in dynamic mode .
half of the films were amorphous and the other half were annealed to the metallic crystalline phase .
it was shown that differences in the gradient of the casimir force between a sphere and an amorphous film on the one hand and between a sphere and a crystalline film on the other hand achieve 20% at separations of about 100 nm . the obtained experimental results were compared with computations using the lifshitz theory and significant deviations were found .
taking into account that according to ref .
@xcite the role of surface roughness was negligible at separations above 70 nm , these deviations might be explained by the influence of localized charges in the amorphous phase and by the use of the drude model in the extrapolation of the optical data in the crystalline phase where aginsbte exhitits metallic conductivity ( see secs . 2 and 3 ) . the possibility to modify the magnitude of the casimir force and even to change attraction for repulsion has long been discussed in the literature .
specifically , in ref .
@xcite the casimir pressure between two magnetodielectric plates was investigated in the approximation of frequency - independent @xmath148 and @xmath149 . for sufficiently large @xmath149 repulsive forces were found .
it was commented in the literature , however , that for real magnetic materials @xmath149 is nearly equal to unity in the range of frequencies , which gives major contribution to the casimir force @xcite . as a result
, the magnitude of @xmath149 always remains far away from the values needed to get the casimir repulsion .
in fact the largest magnitude of @xmath150 is achieved at @xmath151 . for diamagnets @xcite @xmath152 always .
thus , these materials can not be used for modifying the magnitude of the casimir force . for paramagnets
, with the single exception of ferromagnets , the deviation of @xmath153 from unity is also negligibly small at any temperature @xcite .
the subset of paramagnetic materials called _ ferromagnets _ requires special attention when the aim is to modify the casimir force . for such materials @xmath154 at temperatures lower than the so - called _ curie _ temperature @xmath155 @xcite .
the values of @xmath150 for ferromagnets , however , quickly decrease with the increase of frequency .
thus , for ferromagnetic metals
@xmath150 drops toward unity at frequencies of order @xmath156hz and for ferromagnetic dielectrics at frequencies of order @xmath157hz ( note that the first matsubara frequency @xmath158 at @xmath73k is of order @xmath159hz ) . because of this , in all computations using eq .
( [ eq1 ] ) at , say , @xmath160k one can put @xmath161 at all @xmath162 and include ferromagnetic properties only in the zero - frequency term with @xmath6 @xcite .
recently the formalism of functional determinants discussed in sec .
2 was applied to calculate the casimir force acting between an ideal metal cylinder far away from a magnetodielectric plate @xcite .
calculations were performed at @xmath163 .
it was argued that for a cylinder remote sifficiently far from the plate , it is justified to use the static values @xmath164 and @xmath153 for the dielectric permittivity and static permeability of the plate material . under these assumptions ,
both attractive and repulsive casimir force was obtained depending on the values of @xmath164 and @xmath153 .
it should be remarked , however , that in order for magnetic properties to influence the casimir force , the characteristic frequency @xmath165 must be less than @xmath157hz .
hence it follows that cylinder - plate separation must be larger than 15 cm . at such large separations any discussion concerning the sign of the casimir force seems problematic , because this force is equal to zero for all practical purposes and is considerably dominated by newtonian gravity . as was shown in ref .
@xcite , the calculational results for the casimir force between magnetodielectric materials with frequency - dependent @xmath149 depend crucially on whether the zero - temperature or thermal lifshitz formula is used .
the investigation of thermal casimir interaction between two magnetodielectric plates with account of real material properties was performed in refs .
@xcite . as is clear from the foregoing , only ferromagnets may show noticeable effect on the casimir force .
with respect to the casimir effect , however , it is not reasonable to consider parallel plates made of so - called _ hard _ ferromagnetic materials which possess spontaneous magnetization .
the point is that the magnetic interaction between such plates far exceeds any conceivable casimir force .
ferromagnetic materials discussed below are what is referred to as _ soft _ ferromagnetic materials , which do not possess spontaneous magnetization .
it is well known also @xcite that the magnetic permeability of ferromagnets depends on the applied magnetic field ( the so - called _ hysteresis _ ) . keeping in mind that no external magnetic field is applied to the casimir plates and that the mean value of the fluctuating magnetic field is equal to zero , one should consider what is often referred to as _ initial _ permeability , i.e. , @xmath166 .
now we present a few computational results illustrating the impact of ferromagnetic materials on the magnitude and sign of the casimir force .
computations of the casimir pressure were performed using eq .
( [ eq1 ] ) at room temperature with ferromagnetic properties included in the zero - frequency term @xmath6 .
the dielectric properties of ferromagnetic and nonmagnetic metallic plates are described either by the drude model ( [ eq13 ] ) or by the plasma model ( [ eq14 ] ) and the obtained results are compared . keeping in mind that ferromagnetic properties contribute only through the zero - frequency term , it is reasonable to consider not too short separations where the relative contribution of this term is more pronounced .
below we consider separation distances from 0.5 to @xmath167 m , where the role of core electrons in the dielectric response can be neglected .
first we consider the casimir pressure between two plates made of the ferromagnetic metal co with @xmath168 @xcite .
the drude parameters of co are the following @xcite : @xmath169ev and @xmath170ev .
the computational results are presented as a ratio to the zero - temperature casimir pressure between two nonmagnetic plates made of ideal metal at zero temperature @xmath171 in fig .
8 , the solid and dashed lines show the values of @xmath172 as functions of separation computed using the dielectric permittivity ( a ) of the drude model and ( b ) of the plasma model with account of magnetic properties and with magnetic properties disregarded , respectively .
k using ( a ) the drude model and ( b ) the plasma model for the dielectric permittivity .
, width=831 ] quantitatively , the role of magnetic properties can be characterized by the ratio @xmath173 with the increase in separation from 0.5 to @xmath174 m and then to @xmath167 m , @xmath175 computed using the drude model varies from 12% to 44% and to 92% , respectively .
thus , when the drude model is used to describe the dielectric properties of a ferromagnetic metal , the magnetic properties markedly ( up to almost two times at large separations ) increase the magnitude of the casimir pressure @xcite .
as can be seen in fig .
8(b ) , for ferromagnetic metal described by the plasma model the impact of magnetic properties on the casimir pressure is not so pronounced .
quantitatively , from fig .
8(b ) it follows that at separation 0.5 , 2 , and @xmath167 m @xmath176 varies from 6% to 17% , and to 14% , respectively .
mention should be made that if the plasma model is used , the inclusion of magnetic properties decreases the magnitude of the casimir pressure .
similar results were also obtained for fe @xcite . in both cases the inclusion of the magnetic properties does not change the sign of the casimir force leaving it attractive .
very prospective materials for the investigation of the impact of magnetic properties on the casimir force are ferromagnetic dielectrics .
these materials , while displaying physical properties characteristic for dielectrics , demonstrate ferromagnetic behavior under the influence of an external magnetic field @xcite . many ferromagnetic dielectrics are , for instance , composite materials obtained on the basis of a matrix of a polymer compound with inclusion of nanoparticles of ferromagnetic metals . as was shown in ref .
@xcite , the casimir force between two parallel plates made of ferromagnetic dielectric remains attractive , while the influence of magnetic properties on the casimir pressure is much stronger than for co plates considered above . here , we consider the configuration of one plate made of ferromagnetic dielectric ( @xmath177 ) and the other plate made of a nonmagnetic metal au ( @xmath178 ) . as is ahown below , in this configuration
the results for the casimir pressure are drastically different depending on whether the drude or the plasma model is used for the description of au . in the case when the drude model is used , from eqs .
( [ eq3 ] ) and ( [ eq13 ] ) it follows that @xmath179 .
keeping in mind that at zero frequency magnetic properties do not influence the tm reflection coefficient , we arrive at the conclusion that if the drude model is used for the description of au there is no impact of magnetic properties on the casimir pressure between gold and magnetodielectric plates . to perform computations when au plate is described by the plasma model
, one should choose some specific ferromagnetic dielecric .
we consider the composite material on the basis of polystyrene with the volume fraction of ferromagnetic particles in the mixture @xmath180 .
the magnetic permeability of such materials may vary over a wide range @xcite .
below we use @xmath181 .
the dielectric permittivity of polystyrene @xmath182 is presented in the form ( [ eq9 ] ) with @xmath183 .
the parameters of oscillators can be found in ref .
@xcite leading to @xmath184 . as a result
, the dielectric permittivity of the used ferromagnetic dielectric can be presented in the form @xcite @xmath185 the dielectric permittivity of au is described by eq .
( [ eq14 ] ) with @xmath178 , @xmath186ev .
k in the configuration of one plate made of ferromagnetic dielectric and the other plate made of au with account of magnetic properties ( solid line ) and with magnetic properties disregarded ( dashed line ) .
computations are performed using the plasma model for the dielectric permittivity of au .
, width=831 ] the computational results for the relative casimir pressure @xmath187 at @xmath73k as a function of separation are presented in fig .
the solid ( dashed ) lines show the results obtained with magnetic properties of ferromagnetic plate included ( disregarded ) . as is seen in fig .
9 , there is the profound effect of magnetic properties of ferromagnetic dielectric on the casimir pressure .
thus , at separations of 0.5 , 2 , and @xmath167 m the values of @xmath188 are equal to 14% , 60% , and 110% , respectively .
what is more important , the casimir pressure @xmath189 ( and , thus , the casimir force acting between parallel plates through a vacuum gap ) changes sign and becomes positive at separations @xmath190 m .
this means that the casimir force becomes repulsive if au plate is described by the plasma model .
the effect of repulsion between a nonmagnetic metal plate and a ferromagnetic dielectric plate can be used as an experimental test for the influence of magnetic properties on the casimir force and for the model of dielectric permittivity of a metal plate . as was mentioned in the beginning of this section , at the curie temperature
@xmath191 specific for each material , ferromagnets undergo a phase transition @xcite . at higher temperature
they become paramagnets with negligibly small magnetic properties with respect to the casimir force .
the curie temperature varies over a wide temperature region ( for co , for instance , it is equal to 1388k ) . keeping in mind to investigate
the behavior of the casimir pressure under the magnetic phase transition , it is preferable to deal with two plates made of ferromagnetic metal possessing the curie temperature close to room temperature , such as gd ( @xmath191 is of about 290k ) .
dielectric properties of gd are described by either the drude or the plasma model with the parameters @xcite @xmath192ev and @xmath193ev .
the dependence of static magnetic permeability of gd as a function of temperature in the vicinity of curie temperature in modeled in fig .
10(a ) using the data of ref .
@xcite .
m. the solid and dashed lines take into account and disregard the magnetic properties , respectively .
the pairs of lines marked 1 and 2 indicate respective computational results obtained using the drude and plasma models .
, width=831 ] computations of the relative casimir pressure @xmath187 in the configuration of two gd plates were performed at @xmath194 nm as a function of temperature using eq .
( [ eq1 ] ) .
the computational results are presented in fig .
10(b ) , where the pairs of solid and dashed lines marked 1 and 2 indicate the results computed using the drude and plasma models , respectively . as before ,
the solid lines take into account and the dashed lines disregard the magnetic properties of gd . as can be seen in fig .
10(b ) , at @xmath195 the magnetic properties do not influence the casimir pressure , whereas the drude and plasma model approaches lead to results differing for about 19.5% . at @xmath196
the magnetic properties influence the casimir pressure .
quantitatively , at @xmath197k the relative influence of magnetic properties on the casimir pressure is @xmath198% for the drude model and @xmath199% for the plasma model . with account of magnetic properties ,
the relative difference between the predictions of the drude and plasma model approaches for the casimir pressure is approximately equal to 7% .
thus , the magnetic phase transition provides additional opportunities for the invistigation of the impact of magnetic properties on the casimir force and for the selection between different models of dielectric permittivity . to conclude this section
, we note that possible influence of magnetic properties of ferromagnets on the casimir force may be considered as somewhat analogous to the influence of real drift current of conduction electrons discussed in sec
if it is assumed that the fluctuating electromagnetic field can initiate such a current , we arrive at the drude model approach to the thermal casimir force which is considered as most natural by some authors @xcite .
this approach , however , was found to be in contradiction with the results of several experiments @xcite . in this connection
the problem arises whether the fluctuating electromagnetic field can lead to magnetic effects in ferromagnets which remains to be solved .
the most universally known _ normal _ casimir force acts in the direction perpendicular to the interacting surfaces . however ,
when the material surfaces of the interacting bodies are anisotropic or they are positioned asymmetrically , a lateral casimir force may exist , which acts tangential to the surface @xcite .
experimentally the lateral casimir force was first measured @xcite in the configuration of a sinusoidally corrugated au - coated sphere above a sinusoidally corrugated au - coated plate with equal corrugation periods .
the measurement data were compared with theoretical predictions using the pfa .
for two parallel ideal metal plates covered with uniaxial sinusoidal corrugations of equal periods the lateral casimir force was calculated using the exact theory in the second perturbation order with respect to relative corrugation amplitude @xcite . for sinusoidally corrugated plates made of real metals described by the plasma model ,
the lateral casimir force was calculated in refs .
@xcite also in the second - order perturbation theory .
experimentally , the lateral casimir force becomes measurable only for sufficiently large amplitudes of corrugations and at sufficiently short separations .
because of this , calculational methods restricted by the second - order perturbation theory are usually not sufficient for the comparison with the measurement data @xcite . in the first measurement of the lateral casimir force
@xcite the corrugation amplitudes were @xmath200 nm on the plate and @xmath201 nm on the sphere .
the period of corrugations was equal to @xmath202 nm , i.e. , much larger than the separation between the sphere and the plate .
the lateral casimir force with an amplitude of @xmath203n at the shortest separation was found to sinusoidally oscillate as a function of the phase shift @xmath204 between the corrugations .
it was also predicted theoretically , but not observed experimentally due to the use of insufficiently large corrugation amplitudes , that the lateral casimir force is in fact asymmetric , i.e. , deviates from purely sinusoidal harmonic dependence on the phase shift ( this phenomenon is absent when computations are performed up to the second perturbation order ) . in the recently performed new measurements of the lateral casimir force between the au - coated sphere and plate covered with uniaxial sinusoidal corrugations , the corrugation amplitudes were @xmath205 nm on the plate and @xmath206 nm on the sphere ( @xmath207 nm in the second set of measurements ) .
the corrugations had an average period @xmath208 nm , i.e. , more than two times smaller than in refs .
@xcite in order to achieve the regime where the pfa becomes inapplicable to the calculation of the lateral casimir force [ see fig .
11(a , b ) where the corrugated surfaces of the plate and of the sphere used in the first set of measurements , respectively , are shown ] . to obtain corrugations on the sphere with exactly the same period , as on the plate ,
a homogeneous sphere was sandwiched against the plate and pressed by means of a piezo .
in such a manner sinusoidal corrugations from the plate were imprinted on the sphere .
this was done preserving the sphericity of a sphere ( see fig . 12 , where the lighter tone shows higher points on the surface of the sphere used in the first set of measurements ) .
taking into account the sphericity of the second body and sinusoidal corrugations on both bodies in the framework of the pfa and using the lifshitz formula ( [ eq1 ] ) , one arrives to the following approximate expression for the lateral casimir force @xcite @xmath209 . \label{eq28}\ ] ] here , @xmath210 @xmath211 is the bessel function of an imaginary argument , and the following dimensionless variables are used : @xmath212 remembering that eq .
( [ eq28 ] ) uses the pfa for the description of sinusoidal corrugations , it should lead to sufficiently exact results under a condition @xmath213 @xcite . as to the applicability of the pfa to sphere - plate geometry ,
this is garanteed with large safety margin because @xmath214 m .
measurements of the lateral casimir force in refs .
@xcite were performed at separations between the mean levels of corrugations of about 100200 nm . because of this , the application condition of the pfa for the description of corrugations was not satisfied .
therefore , some corrections to eq .
( [ eq28 ] ) due to diffraction - type effects were expected to arise . to take the diffraction - type effects into account ,
the exact description of corrugations using the rayleigh scattering theory was developed @xcite while preserving the use of the pfa for the description of sphere - plate geometry , where it is sufficiently accurate .
as a result , the following exact " expression for the lateral casimir force was obtained @xcite @xmath215 .
\nonumber\end{aligned}\ ] ] here , @xmath216 ( @xmath217 ) is the reflection matrix of the downward ( upward ) moving waves on the bottom ( top ) grating under a condition that the top ( bottom ) grating is removed , and @xmath218 is an infinite dimensional unit matrix . in both eqs .
( [ eq28 ] ) and ( [ eq31 ] ) the dielectric properties of au were taken into account by means of the generalized plasma - like model ( [ eq21 ] ) . in fig .
13(a ) we show the measured data in the first set of measurements ( dots ) and computed using the exact theory of eq .
( [ eq31 ] ) lateral casimir force ( solid line ) over four corrugation periods versus the normalized lateral displacement @xmath219 between the corrugated surfaces at a separation @xmath220 nm ( similar figures related to larger separations and to the second set of measurements can be found in ref .
@xcite ) . as can be seen in fig .
13(a ) , the lateral casimir force is a periodic function of the phase shift .
the important characteristic feature of the periodic curve shown in the figure is that it is _ asymmetric _ , i.e. , the dependence of @xmath221 on @xmath219 is not strictly sinusoidal ( note that the sinusoidal dependence of the lateral force on @xmath222 holds only if the calculations are restricted to the second order in corrugation amplitudes @xmath223 ) . the true asymmetry of the measured lateral force is obvious even without the theory line .
for example , in fig .
13(a ) the average shift of each maximum point from the midpoint of two adjacent minima is @xmath224 .
we emphasize that the theoretical solid line was obtained with no fitting parameters .
as can be seen in fig .
13(a ) , the exact theory of eq .
( [ eq31 ] ) is in a very good agreement with the experimental data shown as dots .
the experimental data for the @xmath225 versus separation @xmath0 are shown in fig .
13(b ) as crosses .
the arms of crosses indicate the total experimental error determined at a 95% confidence level for each experimental point individually .
the exact computational results for the @xmath225 are shown in fig .
13(b ) as a band between the two solid lines versus separation .
the width of this band takes into account the computational errors and the correction to eq .
( [ eq31 ] ) due to surface roughness . as is seen in fig .
13(b ) , the exact theory is in a very good agreement with the measurement data .
the computational results for the @xmath225 at different separations obtained using the pfa [ eq . ( [ eq28 ] ) ] are shown in fig .
13(b ) as a band between two dashed lines .
the width of this band takes into account the role of surface roughness . as can be seen in fig .
13(b ) , the experimental data indicated as crosses are inconsistent with the prediction of the pfa approach , thus revealing the role of diffraction - type effects for corrugations of relatively small period used in this experiment .
besides fundamental interest in the phenomenon of lateral casimir force as one more manifestation of the zero - point oscillations of electromagnetic field , it is pertinent to note that this version of the casimir force gives the possibility to actuate lateral translations of corrugated surfaces in micromachines .
the lateral casimir force might be used to solve the tribological problems plaguing the microdevice industry .
specifically , it was proposed to use this effect for the frictionless transmission of lateral motion by means of a nanoscale rack and pinion without intermeshing cogs @xcite or in a ratchet with asymmetric corrugations driven by the casimir force @xcite .
such devices could be used for transfering motion between two microelectromechanical systems without mechanical contact .
the experimental and theoretical results on the lateral casimir force discussed above bring us closer to the realization of micromachines driven by the zero - point fluctuations of quantum vacuum .
in the foregoing , we have listed several experiments and theoretical advances in the casimir effect which has come to the center of public attention very recently .
the casimir effect deals with quantum vacuum , which is the most fundamental and puzzling type of physical reality .
this is a possible reason why the research area in this field is so wide . on the one hand , it is closely connected with foundations of quantum statistical physics and even calls for reconsideration of some of its basic concepts . on the other hand , the casimir effect quickly turns into a practical tool of nanotechnology which is expected to find numerous applications in the next generation of microdevices . as was discussed in sec . 2 , the fundamental lifshitz theory of the van der waals and casimir forces comes into conflict with basics of thermodynamics when the relaxation properties of free charge carriers are taken into account . in sec . 3 ,
the three landmark recently performed experiments are described demonstrating that in a literal sense , the lifshitz theory is in a deep contradiction with the measurement data .
only if we accept the phenomenological prescription that for dielectrics charge carriers are disregarded and for metals they are described by the plasma model , the theoretical predictions of the lifshitz theory come in agreement with the data .
this situation can not be considered as satisfactory .
in fact we deal with some kind of a paradox .
the mathematical formalism of the lifshitz theory during the last few years has undergone far reaching generalization discussed in secs . 2 and 5 .
it is presently possible to write the lifshitz - type formulas for compact objects of arbitrary shape in terms of reflection amplitudes of electromagnetic oscillations on their surfaces .
based on this it was even believed that at the present time computation of the casimir force between any bodies is only a problem of sufficient computer facilities . as was argued in secs . 2 and 3
, the problem is , however , much deeper .
in fact the reflection amplitudes suggested by our knowledge of physical processes with real ( not fluctuating ) electromagnetic fields lead to computational results in contradiction with the experimental data making powerful mathematical formalism at least insufficient .
this is a fundamental problem to be solved in the future . from the point of view of applications in nanotechnology , large progress on the control the magnitude of the casimir force
was achieved .
this was done by means of a modification of dielectric properties by using different materials and different influential factors such as irradiation with laser light . in several experiments considered in sec .
4.1 it was possible to modify the magnitude of the casimir force from a few percent to about 50% .
there were several proposals on how to achieve the casimir repulsion through the vacuum gap , but to the present day the casimir repulsion was realized only in the familiar three - layer system .
one more opportunity to control the magnitude of the casimir force is suggested by phase transitions . in sec .
4.1 we have discussed the phase transition from dielectric to the metallic state .
a few tens of percent variation in the magnitude of the casimir force is achievable under a phase transition of the plate material .
quick progress in the application of the casimir force in nanotechnology might be caused by the use of ferromagnetic materials .
for real material properties this subject was investigated only recently ( see sec .
what is most striking , according to the lifshitz theory the casimir interaction of a nonmagnetic metallic plate described by the plasma model with a ferromagnetic dielectric plate through the vacuum gap becomes repulsive under some conditions .
this effect , if exists in nature , opens up the way for the developments of various microdevices .
we should notice , however , that the influence of magnetic properties on the casimir interaction is not yet confirmed experimentally .
the confirmation or exclusion of such an influence is expected in the near future .
one more opportunity for applications in nanotechnology is suggested by the lateral casimir force discussed in sec .
5 . together with the normal casimir force
, it gives the possibility to actuate both normal and lateral translations of corrugated surfaces in micromachines by means of the electromagnetic zero - point oscillations .
with respect to tribological problems , the lateral casimir force allows frictionless transition of lateral motion .
all the above allows to conclude that the casimir effect promises important breakthroughs in both fundamental physics and its applications in nanotechnology .
99 [ 1 ] casimir h b g 1948 _ proc .
_ * 51 * 793 [ 2 ] buks e and roukes m l 2001 _ phys . rev .
_ b * 63 * 033402 [ 3 ] chan h b , aksyuk v a , kleiman r n , bishop d j and capasso f 2001 _ science _ * 291 * 1941 [ 4 ] lifshitz e m 1956 _ zh . eksp .
_ * 29 * 94 ( _ sov . phys
* 2 * 73 ) [ 5 ] dzyaloshinskii i e , lifshitz e m and pitaevskii l p 1961 _ sov .
_ * 4 * 153 [ 6 ] klimchitskaya g l , mohideen u and mostepanenko v m 2009 _ rev .
phys . _ * 81 * 1827 [ 7 ] mostepanenko v
m and klimchitskaya g l 2010 _ int .
phys . _ a * 25 * 2302 [ 7a ] bordag m , klimchitskaya g l , mohideen u and mostepanenko v m 2009 _ advances in the casimir effect _
( oxford : oxford university press ) [ 8 ] decca r s , fischbach e , klimchitskaya g l , krause d e , lpez d and mostepanenko v m 2003 _ phys .
_ d * 68 * 116003 [ 9 ] decca r s , lpez d , fischbach e , klimchitskaya
g l , krause d e and mostepanenko v m 2005 _ ann .
ny _ * 318 *
37 [ 10 ] decca r s , lpez d , fischbach e , klimchitskaya g l , krause d e and mostepanenko v m 2007 _ phys .
_ d * 75 * 077101 [ 11 ] decca r s , lpez d , fischbach e , klimchitskaya g l , krause d e and mostepanenko v m 2007 _ eur .
c * 51 * 963 [ 12 ] chen f , klimchitskaya g l , mostepanenko v m and mohideen u 2007 _ optics express _ * 15 * 4823 [ 13 ] chen f , klimchitskaya g l , mostepanenko v m and mohideen u 2007 _ phys .
_ b * 76 * 035338 [ 14 ] obrecht j m , wild r j , antezza m , pitaevskii l p , stringari s and cornell e a 2007 _ phys .
lett . _ * 98 * 063201 [ 15 ] klimchitskaya g l and mostepanenko v m 2008 _ j. phys . a : math .
theor . _ * 41 * 312002 [ 16 ] golestanian r and kardar m 1997 _ phys .
_ * 78 * 3421 [ 17 ] chen f , mohideen u , klimchitskaya g l and mostepanenko v m 2002 _ phys .
lett . _ * 88 * 101801 [ 18 ] chen f , mohideen u , klimchitskaya g l and mostepanenko v m 2002 _ phys
. rev . _ a * 66 * 032113 [ 19 ] chiu h - c , klimchitskaya g l , marachevsky v n , mostepanenko v m and mohideen u 2009 _ phys .
_ b * 80 * 121402(r ) [ 20 ] chiu h - c , klimchitskaya g l , marachevsky v n , mostepanenko v m and mohideen u 2010 _ phys .
_ b * 81 * , 115417 [ 21 ] bimonte g , klimchitskaya g l and mostepanenko v m 2009 _ phys .
rev . _ a * 79 * 042906 [ 22 ] zhou f and spruch l 1995 _ phys .
rev . _ a * 52 * 297 [ 23 ] toma m s 2002 _ phys .
rev . _ a * 66 * 052103 [ 24 ] raabe c , knll l and welsch d - g 2003 _ phys .
rev . _ a * 68 * 033810 [ 25 ] emig t , jaffe r l , kardar m and scardicchio a 2006 _ phys .
lett . _ * 96 * 080403 [ 26 ] bulgac a , magierski p and wirzba a 2006 _ phys .
_ d * 73 * 025007 [ 27 ] kenneth o and klich i 2006 _ phys .
lett . _ * 97 * 160401 [ 28 ] bordag m 2006 _ phys .
_ d * 73 * 125018 [ 29 ] kenneth o and klich i 2008 _ phys . rev .
_ b * 78 * 014103 [ 29a ] bordag m and pirozhenko i 2010 _ phys .
_ d * 81 * 085023 [ 30 ] wilson a g 1931 _ proc .
lond . _ a * 133 * 458 [ 31 ] mott n f 1990 _ metal - insulator transitions _ ( london : taylor and francis ) [ 32 ] parsegian v a 2005 _ van der waals forces : a handbook for biologists , chemists , engineers , and physicists _ ( cambridge : cambridge university press ) [ 32a ] inui n 2003 j. phys .
. jap . * 72 * 2198 [ 33 ] palik e d ( ed ) 1985 _ handbook of optical constants of solids _
( new york : academic ) [ 34 ] geyer b , klimchitskaya g l and mostepanenko v m 2005 _ phys .
_ d * 72 * 085009 [ 35 ] klimchitskaya g l and geyer b 2008 _ j. phys .
* 41 * 164032 [ 36 ] bezerra v b , klimchitskaya g l and mostepanenko v m 2002 _ phys .
_ a * 66 * 062112 [ 37 ] bezerra v b , klimchitskaya g l , mostepanenko v m and romero c 2004 _ phys .
rev . _ a * 69 * 022119 [ 38 ] bostrm m and sernelius b e 2004 _ physica _ a * 339 * 53 [ 39 ] hye j s , brevik i , ellingsen s a and aarseth j b 2007 _ phys .
_ e * 75 * 051127 [ 40 ] brevik i , ellingsen s a , hye j s and milton k a 2008 _ j. phys . a : math .
theor . _ * 41 * 164017 [ 41 ] klimchitskaya
g l and mostepanenko v m 2008 _ phys .
_ e * 77 * 023101 [ 42 ] klimchitskaya g l , mohideen u and mostepanenko v m 2008 _ j. phys . a : math .
theor . _ * 41 * 432001 [ 43 ] pitaevskii l p 2008 _ phys .
_ * 101 * 163202 [ 44 ] dalvit d a r and lamoreaux s k 2008 _ phys .
* 101 * 163203 [ 45 ] chazalviel j - n 1999 _ coulomb screening of mobile charges : applications to material science , chemistry and biology _ ( boston : birkhauser ) [ 46 ] geyer b , klimchitskaya g l , mohideen u and mostepanenko v m 2009 _ phys .
lett . _ * 102 * 189301 [ 47 ] mostepanenko v m , decca r s , fischbach e , geyer b , klimchitskaya g l , krause d e , lpez d and mohideen u 2009 _ int . j. mod .
phys . _ a * 24 * 1721 [ 47a ] klimchitskaya g l 2009 _ j. phys .
ser . _ * 161 * 012002 [ 47b ] dure j c , maass p , roling b and sidebottom d l 2009 _ rep .
_ * 72 * 046501 [ 47c ] shklovskii b i and efros a l 1984 _ electronic properties of doped semiconductors .
solid state series _
, v.45 ( berlin : springer ) [ 48 ] decca r s , fischbach e , geyer b , klimchitskaya g l , krause d e , lpez d , mohideen u and mostepanenko v m 2009 _ phys .
lett . _ * 102 * 189303 [ 48a ] mostepanenko v m , 2009 _ j. phys .
ser . _ * 161 * 012003 [ 49 ] pitaevskii l p 2009 _ phys . rev
. lett . _ * 102 * 163203 [ 50 ] dalvit d a r and lamoreaux s k 2009 _ phys .
* 102 * 163204 [ 51 ] krause d e , decca r s , lpez d and fischbach e 2007 _ phys . rev .
lett . _ * 98 * 050403 [ 52 ] canaguier - durand a , maia neto p a , lambrecht a and reynaud s 2010 _ phys .
lett . _ * 104 * 040403 [ 52a ] canaguier - durand a , maia neto p a , lambrecht a and reynaud s 2010 _ phys .
_ a * 82 * 012511 [ 53 ] zandi r , emig t and mohideen u 2010 _ phys .
_ b * 81 * 195423 [ 54 ] lambrecht a and reynaud s 2000 _ eur .
d * 8 * 309 [ 55 ] pirozhenko i , lambrecht a and svetovoy v b 2006 _ new j. phys . _
* 8 * 238 [ 56 ] svetovoy v b , van zwol p j , palasantzas g and hosson j th m 2008 _ phys .
rev . _ b * 77 * 035439 [ 57 ] decca r s , lpez d and osquiguil e 2010 _ int . j. mod .
a * 25 * 2223 [ 57a ] geyer b , klimchitskaya g l and mostepanenko v m 2010 _ phys . rev .
_ b * 81 * 245421 [ 58 ] inui n 2004 j. phys .
* 73 * 332 [ 59 ] inui n 2006 j. phys .
* 75 * 024004 [ 59a ] inui n 2007 j. phys .
ser . * 89 * 012018 [ 60 ] mostepanenko v
m and geyer b 2008 _ j. phys . a : mat .
* 41 * 164014 [ 69a ] shapiro b 2010 _ phys .
_ b * 82 * 075205 [ 62 ] antezza m , pitaevskii l p and stringari s 2004 _ phys .
rev . _ a * 70 * 053619 [ 63 ] antezza m , pitaevskii l p and stringari s 2005 _ phys .
lett . _ * 95 * 113202 [ 64 ] harber d m , obrecht j m , mcguirk j m and cornell e a 2005 _ phys .
rev . _ a * 72 * 033610 [ 65 ] caride a o , klimchitskaya g l , mostepanenko v m and zanette s i 2005 _ phys .
rev . _ a * 71 * 042901 [ meta1 ] silveirinha m g 2010 _ phys .
_ b * 82 * 085101 [ 66 ] chen f , mohideen u , klimchitskaya g l and mostepanenko v m 2005 _ phys .
_ a * 72 * 020101(r ) [ 67 ] chen f , mohideen u , klimchitskaya g l and mostepanenko v m 2006 _ phys .
rev . _ a * 74 * 022103 [ 68 ] chen f , klimchitskaya g l , mostepanenko v m and mohideen u 2006 _ phys .
_ * 97 * 170402 [ 69 ] deman s , heeck k , wijngaarden r j and iannuzzi d 2009 _ phys .
* 103 * 040402 [ 70 ] milling a , mulvaney p and larson i 1996 _ j. colloid interface sci . _ * 180 * 460 [ 71 ] meurk a , luckham p f and bergstrm l 1997 _ langmuir _ * 13 * 3896 [ 72 ] lee s - w and sigmund m w 2002 _ colloids surf . _ a * 204 * 43 [ 73 ] munday j n , capasso f and parsegian v a 2009 _ nature _ * 457 * 170 [ 74 ] geyer b , klimchitskaya g l , mohideen u and mostepanenko v m 2008 _ phys . rev . _ a * 77 * 036102 [ 74a ] van zwol p j , palasantzas g and hosson j th m 2009 _ phys .
_ b * 79 * 195428 [ 75 ] klimchitskaya g l , mohideen u and mostepanenko v m 2007 _ j. phys .
_ * 40 * f841 [ 76 ] mostepanenko v
m and trunov n n 1997 _ the casimir effect and its applications _ ( oxford : oxford university press ) [ 77 ] bimonte g , calloni e , exposito g , milano l and rosa l 2005 _ phys .
lett . _ * 94 * 180402 [ 78 ] bimonte g , calloni e , exposito g and rosa l 2005 _ nucl .
_ b * 726 * 441 [ 79 ] bimonte g 2008 _ phys .
rev . _ a * 78 * 062101 [ 80 ] castillo - garza r , chang c - c , jimenez d , klimchitskaya g l , mostepanenko v m and mohideen u 2007 _ phys .
rev . _ a * 75 * 062114 [ 81 ] pirozhenko i and lambrecht a 2008 _ phys .
rev . _ a * 77 * 013811 [ 82 ] torricelli g , van zwol p j , shpak o , binus c , palasantzas g , kooi b j , svetovoy v b and wuttig m 2010 _ phys .
rev . _ a * 82 * 010101(r ) [ 83 ] kenneth o , klich i , mann a and revzen m 2002 _ phys .
lett . _ * 89 * 033001 [ 84 ] iannuzzi d and capasso f 2003 _ phys .
lett . _ * 91 * 029101 [ 85 ] vonsovskii s v 1974 _ magnetism _ ( new york : j. wiley ) .
[ 86 ] geyer b , klimchitskaya g l and mostepanenko v m 2010 _ phys .
_ b * 81 * 104101 [ 87 ] klimchitskaya g l , geyer b and mostepanenko v m 2010 _ int . j. mod .
phys . _ a * 25 * 229360 [ 88 ] rahi s j , emig t , graham n , jaffe r l and kardar m 2009 _ phys .
_ d * 80 * 085021 [ 89 ] toma m s 2005 _ phys .
lett . _ a * 342 * 381 [ 90 ] goldman a 1999 _ handbook of modern ferromagnetic materials _ ( new york : springer ) [ 91 ] ordal m a , bell r j , alexander jr r w , long ll and querry m r 1985 _ appl .
_ * 24 * 4493 [ 92 ] chambers s a 2006 _ surf .
rep . _ * 61 * 345 [ 93 ] roy k and datta a n 1974 _ j. phys .
_ d * 7 * 1053 [ 95 ] lewin l 1947 _ j. inst .
_ part 3 * 94 * 64 [ 96 ] ausloos m ( ed ) 1983 _ magnetic phase transitions _
( new york : springer ) [ 97 ] palik e d ( ed ) 1998 _ handbook of optical constants of solids iii _
( new york : academic ) [ 98 ] coey j m d , skumryev v and gallagher k 1999 _ nature _ * 401 * 35 [ 99 ] emig t , hanke a , golestanian r and kardar m 2003 _ phys .
_ a * 67 * 022114 [ 100 ] rodrigues r b , maia neto p a , lambrecht a and reynaud s 2006 _ phys .
lett . _ * 96 * 100402 [ 101 ] rodrigues r b , maia neto p a , lambrecht a and reynaud s 2006 _ phys .
rev . _ a * 75 * 062108 [ 102 ] chen f , mohideen u , klimchitskaya g l and mostepanenko v m 2006 _ phys .
_ * 98 * 068901 [ 103 ] ashourvan a , miri m and golestanian r 2007 _ phys .
lett . _ * 98 * 140801 [ 104 ] ashourvan a , miri m and golestanian r 2007 _ phys .
_ e * 75 * 040103(r ) [ 104a ] ashourvan a , miri m and golestanian r 2007 _ j. phys .
ser . _ * 89 * 012017 [ 105 ] emig t 2007 _ phys .
_ * 98 * 160801 | we discuss recent experimental and theoretical results on the casimir force between real material bodies made of different materials .
special attention is paid to calculations of the normal casimir force acting perpendicular to the surface with the help of the lifshitz theory taking into account the role of free charge carriers .
theoretical results for the thermal casimir force acting between metallic , dielectric and semiconductor materials are presented and compared with available experimental data .
main attention is concentrated on the possibility to control the magnitude and sign of the casimir force for applications in nanotechnology . in this respect
we consider experiments on the optical modulation of the casimir force between metal and semiconductor test bodies with laser light .
another option is the use of ferromagnetic materials , specifically , ferromagnetic dielectrics . under some conditions this allows to get casimir repulsion .
the lateral casimir force acting between sinusoidally corrugated surfaces can be considered as some kind of noncontact friction caused by zero - point oscillations of the electromagnetic field .
recent experiments and computations using the exact theory have demonstrated the role of diffraction - type effects in this phenomenon and the possibility to get asymmetric force profiles .
conclusion is made that the casimir force may play important role in the operation of different devices on the nanoscale . |
the assumption of conformal invariance in two - dimensional critical systems has been very fruitful @xcite .
exact expressions for correlation functions , as well as values of critical exponents for minimal models , are among the most stringent predictions of conformal theory .
recently , the interest has been focused on fractal curves that may be realized as interfaces induced in a critical conformal - invariant system by imposing symmetry - breaking boundary conditions @xcite .
it is believed that the geometric properties of this kind of curves can be described by a family of self - avoiding stochastic processes introduced by schramm .
the self - avoidance property is obtained by iteratively removing , during the growth of the curve , the hull formed from the curve itself and all non - reachable domains enclosed by the curve .
this removal is performed by a conformal transformation . as a consequence ,
the construction relies on both the assumption of conformal invariance and on markovian property .
the family of curves obtained in this way is parameterized by a single parameter @xmath1 corresponding to the intensity of the noise in the underlying brownian process .
+ in a conformal field theory characterized by a central charge @xmath5 , it has been predicted that the interfaces induced by symmetry - breaking boundary conditions are described by the chordal schramm - lwner process of parameter @xmath1 with @xmath6 the length @xmath7 of the interface is expected to scale anomalously with the lattice size as @xmath8 where the fractal dimension is related to @xmath1 by @xmath9 the probability that a given point be on the left of the interface is given by the schramm formula : @xmath10 this formula is valid in the upper half - plane @xmath11 with fixed boundary conditions on the real axis , changing for @xmath12 .
it can easily be extended to other geometries by taking advantage of the conformal symmetry of the critical system that allows to apply a conformal transformation from the upper - half plane to another geometry . in the following
, we will consider a critical system confined in a square .
the mapping of the point @xmath13 in upper - half plane to the point @xmath14 in the square is realized by the schwartz - christoffel conformal transformation @xmath15 where @xmath16 is the incomplete elliptic integral , @xmath17 is the complete elliptic integral and @xmath18 is related to the aspect ratio @xmath19 of the square by @xmath20 where @xmath21 is the associated elliptic integral .
+ the critical properties of the @xmath22-state potts model ( @xmath23 ) defined by the hamiltonian @xcite @xmath24 are described by the conformal theory with central charge @xmath25 where the integer @xmath26 is related to the number of states @xmath22 by the nienhuis formula @xmath27 one thus expects the geometric properties of the interfaces induced in the potts model to be described in the scaling limit by the chordal @xmath4 with the two possible values @xmath28 this prediction has been proved exactly by smirnov in the case of percolation @xcite ( @xmath29 ) and latter for the ising model @xcite ( @xmath30 in the fortuin - kasteleyn representation . in the case of the @xmath31 potts model , the fractal dimension of the interfaces between spin clusters has been shown to be compatible within numerical accuracy with equation ( [ defdf ] ) and @xmath32 @xcite .
predictions of @xmath4 have been tested more specifically for this model by gamsa and cardy @xcite .
these authors considered the interfaces between the spin clusters and between the fortuin - kastelyn clusters and they extracted the parameter @xmath1 both from the scaling of the length of the interface and from the schramm formula for the left - passage probability .
several geometries and boundary conditions were considered .
as expected since @xmath33 , gamsa and cardy measured the parameter @xmath32 for the spin interfaces and @xmath34 for the interfaces between the fortuin - kastelyn clusters , in agreement with the values expected for @xmath33 model . unlike in the ising model , certain ambiguities arise for the interfaces in the potts model . when fixing two different potts states on the boundaries at the left and at the right of the system ,
two interfaces are usually induced since a cluster of the third state may appear between the left and right clusters .
gamsa and cardy generalized the schramm formula for the left - passage probability to the case of two sle curves @xcite .
unfortunately , in the 3-state potts model , the two curves sometimes collapse in a finite system and the latter formula for the left - passage probability does not hold . to circumvent this problem , gamsa and cardy
have also studied fluctuating boundary conditions : the left boundary is fixed to one of the potts states while at the right boundary , spins are free to take any of the two other potts states . with these boundary conditions ,
only one interface is induced and the original schramm formula for the left - passage probability can be applied .
recently , the fractal dimension of the interface has been numerically shown to be compatible with @xmath35 even for non - integer values of the number of states @xmath22 @xcite .
+ the ferromagnetic random - bond 3-state potts model is defined by the hamiltonian @xmath36 when the couplings are distributed according to the binary distribution @xmath37\ ] ] the model is critical on the self - dual line @xmath38 randomness is a relevant perturbation and thus alters the critical behavior . in numerical simulations ,
the ratio @xmath39 is usually fixed to the value that minimizes the scaling corrections due to the cross - over with the pure and percolation fixed points .
the optimum @xmath40 can be determined as the value leading to the largest central charge extracted from the scaling of the free energy density on the strip .
even though conformal symmetry is broken for a single disorder realization , it is restored after averaging over randomness .
magnetization profiles or correlation functions in confined geometries have indeed been shown to transform covariantly under conformal transformations @xcite .
recently , the interface induced by boundaries has been studied in the perspective of @xmath4 @xcite .
the scaling of the interface length has been determined both by transfer matrix techniques and monte carlo simulations . from the fractal dimension @xmath0 , jacobsen _ et al . _
estimated @xmath41 for the spin interface .
duality , i.e. @xmath42 , is satisfied within error bars .
the results are compatible with renormalization group arguments .
note that this estimate of @xmath43 is not compatible with the value @xmath44 given by relation ( [ eq1 ] ) with the estimate of the central charge @xmath45 .
+ the schramm formula has never been tested in the random potts model .
it has already been considered in other random systems : ising spin - glasses @xcite and the solid - on - solid model on a random substrate @xcite . in the second case , a large discrepancy was observed between the two estimates of @xmath1 obtained from the scaling of the interface length and from the left - passage probability .
our aim is to check whether such a discrepancy exists for the random - bond potts model too .
we will restrict ourselves to a square geometry with fluctuating boundary conditions and to interfaces between spin clusters . in the first section , we present the numerical details of the monte carlo simulations that we have performed .
then , the fractal dimension of the interfaces is studied with an emphasis on the influence of disorder on the value of @xmath1 . in the last section
, we present our results for the left - passage probability and the estimation of @xmath1 that can be made from it .
we have studied the random - bond 3-state potts model using large - scale monte carlo simulations .
square lattices were considered with fluctuating boundary conditions and sizes @xmath46 , 90 , 128 , 180 , 256 , 360 and 512 .
each monte carlo step consists into one iteration of the metropolis algorithm and one of the swendsen - wang algorithm @xcite .
this choice is motivated by the fact that the dynamics of an interface with the swendsen - wang algorithm is very slow and may even be slower than with local monte carlo algorithms ( see for example table 1 of @xcite ) . the combination of the two algorithms may help the system to escape dynamical traps . for equilibration of the system , between @xmath47 ( @xmath46 ) and @xmath48 ( @xmath49 ) monte carlo steps were discarded ( table [ table1 ] ) .
measurements were taken every five monte carlo steps .
between @xmath50 ( @xmath46 ) and @xmath51 ( @xmath49 ) measurements were made for each random realization .
these values were chosen in order to achieve a good thermalization and then a sufficient sampling of the equilibrium probability distribution .
the number of measurements may appear at first sight unnecessarily large since it may be argued that the average over randomness will smooth thermal fluctuations .
smaller numbers of measurements have indeed led to accurate values of the critical exponents in the past . however , in our problem , preliminary tests have shown that a smaller number of monte carlo steps leads to systematic deviations on the estimate of @xmath1 due to the large autocorrelation time of the interface .
+ the measurements were then averaged over @xmath52 different random configurations of the exchange couplings @xmath53 with the optimal ratio @xmath54 determined by jacobsen _
et al . _
since the random realizations are uncorrelated , the measurements obtained for each of them are statistically independent random variables so that the error can be estimated , according to the central - limit theorem , as @xmath55 where @xmath56 is the mean square deviation of the observable among the random configurations .
additional simulations have been performed for other values @xmath57 of the amplitude of randomness .
the same number of monte carlo steps were used but only 3000 random realizations were produced and the lattices sizes were limited to @xmath46 , 90 and 128 .
@*3l @xmath58 & thermalization & measurement + 64 & 25000 & 75000 + 90 & 28750 & 86250 + 128 & 33060 & 99185 + 180 & 38015 & 114060 + 256 & 43715 & 131165 + 360 & 50270 & 150835 + 512 & 57810 & 173460 + for a given spin configuration , the interface is determined by applying a simple algorithm : starting from the upper point of the boundary where the conditions change and thus the interface is fixed , the interface is built on the dual lattice by going through the broken bonds between neighboring spins ( see figure [ fig0 ] for an example ) .
the interface goes straight as long as the spin at the right is in the same state as the left boundary and the one at the left in one of the two other states .
ambiguous situations may arise : when the interface can not go straight , the spin at the right of the interface can be in some cases in the same state as the left boundary whether the interface turns on the right or on the left .
the algorithm is thus not able to choose one of the two possible directions . on figure [ fig0 ] , four of these situations
are depicted .
we arbitrarily imposed the rule that the interface should always turn on the right in these cases ( _ turn - right tie - breaking _ rule ) .
as a consequence , the algorithm may fail to detect the correct interface .
because of the rule of always turning on the right in case of ambiguity , the interface may touch itself and thus form loops enclosing a cluster in a different state . as shown on figure [ fig0 ] , this can only happen on the left of the interface .
direct visualization of spin configurations shows that the number of these loops is of order @xmath59 and that the cluster sizes are most of the time limited to a few spins and are always much smaller than the lattice size . + in order to gauge the influence of these unwanted loops on the scaling dimension of the interface and on the left - passage probability , we estimated @xmath1 for the pure 3-state model and compared with the expected value @xmath60 .
as will be seen in the following , the scaling dimension is not affected by the presence of these loops .
we note that the absence of noticeable effect on the scaling dimension has also been reported recently in the case of the pure ising model @xcite .
it is not the case however for the left - passage probability .
a small but systematic deviation from the schramm formula is observed on the left part of the lattice .
this deviation affects the estimation of @xmath1 . following the procedure presented in section [ sec3 ] ,
we obtain @xmath61 for @xmath49 .
since the loops appears only on the left of the interface , it creates a left - right asymmetry .
fortunately , the restriction of the interpolation of the data with the schramm formula to the right third of the lattice , i.e. for @xmath62\times [ 0;l-1]$ ] only turns out to give @xmath63 for @xmath49 in much better agreement with the expected value @xmath60 for the pure model .
the scaling of the interface length was already determined by jacobsen _
et al._. besides reproducing their result , we present here a study of the influence of disorder .
as already mentioned , we will restrict ourselves to the interfaces between spin clusters .
for the pure model , we obtain @xmath64 corresponding to @xmath65 , in very good agreement with the expected value @xmath60 .
our data for the disorder amplitude @xmath66 , i.e. at the random fixed point , is plotted on figure [ fig3 ] .
a simple power - law fit @xmath8 leads to the fractal dimension @xmath67 and thus to the value @xmath68 .
however , a deviation from a pure power - law can be observed , for example , by removing the smallest lattices sizes . in the inset of figure [ fig3 ] ,
the fractal dimension obtained when taking into account in the fit only the data with lattice sizes larger than @xmath58 is plotted with respect to @xmath69 .
while the effective fractal dimension for the pure system is relatively stable , a tendency to lower values of @xmath0 is clearly observed in the random case .
the last points are very noisy because the interpolation is made only with the two or three largest lattice sizes .
the extrapolation to the thermodynamic limit is then delicate .
should the two last points be interpreted as a real trend or a statistical fluctuation ?
our final estimate will be @xmath70 which correspond to @xmath2 .
this value is less accurate but compatible within error bars with the estimate given by jacobsen _
( @xmath71 ) .
the fractal dimension @xmath0 and thus @xmath1 are sensitive to the estimate @xmath40 of the disorder amplitude at the random fixed point . as shown on figure [ fig2 ] , @xmath0 and thus @xmath1 diminishes when the disorder gets stronger .
a linear interpolation of @xmath0 versus @xmath57 allows for a rough estimation of the additional error @xmath72 due to the uncertainty @xmath73 on @xmath40 . with the value @xmath74 given by jacobsen
, we obtained @xmath75 and thus @xmath76 , an order of magnitude smaller than the statistical error . besides the necessity to average over random realizations , numerical simulations of random systems present the difficulty that an observable @xmath77 may not be self - averaging .
in such a situation , the average of @xmath77 over @xmath78 disorder realizations goes to @xmath79 in the limit @xmath80 but the relative fluctuations @xmath81/\langle\phi^2\rangle$ ] do not vanish .
the probability distribution of the length @xmath7 of the interface is shown on figure [ fig1 ] for different lattice sizes @xmath58 . when plotted with respect to @xmath82
, the probability distribution appears the same for all lattice sizes .
it means that the mean square deviation @xmath83 scales as @xmath84 .
the cumulant @xmath85 is thus constant : the interface length is non self - averaging .
another important feature of random systems is that the typical event may differ from the average .
as seen on figure [ fig1 ] , the probability distribution of the interface length is not symmetric and thus these events indeed differ slightly .
however , the distribution does not present any long tail with rare events difficult to observe in numerical simulations .
we rather observe a distribution close to a gaussian .
finally , random systems have the peculiar property to possibly display multifractality : different moments of an observable @xmath77 may present a different scaling behavior , i.e. @xmath86 where @xmath87 depends of the order @xmath88 of the moment .
the first moments @xmath89 have been computed .
all fractal dimensions @xmath90 corresponding to the scaling behavior @xmath91 are compatible within error bars .
we conclude that the interface length is not multifractal .
we now determine @xmath1 from the schramm formula of the left - passage probability .
we follow the same procedure as @xcite .
the left - passage probability @xmath92 was estimated numerically for all points @xmath93 of the lattice . to extract @xmath1 , the square deviation from the schramm formula @xmath94 at any point of the lattice @xmath95
^ 2\ ] ] was computed and then averaged only in the right third of the lattice : @xmath96 the value of @xmath1 leading to the smallest mean square deviation is then searched as follows : the mean square deviation @xmath97 is first computed for a set of values @xmath1 and then interpolated with a 4th - order polynomial using a weight @xmath98 in order to give a higher weight to the points in the neighborhood of the minimum .
the latter is then determined numerically by searching for the root of the derivative of the 4th - order polynomial .
an example of this procedure is shown on figure [ fig4 ] .
for the largest lattice size @xmath49 , the mean square deviation is minimal for @xmath99 .
the monitoring of @xmath100 shows that for this value of @xmath1 , the square deviation is of order @xmath101 in the right third of the lattice ( figure [ fig4a ] ) .
the square of the statistical error @xmath102 is of the same order . the left - right asymmetry due to the loops
is clearly seen : @xmath103 is larger by two orders of magnitude in the left part of the lattice .
the much better agreement of the data with the schramm formula in the right third of the lattice justifies _ a - posteriori _ the restriction of the interpolation to this region . on figure [ fig5 ] ( left ) , @xmath1 is plotted versus the number of disorder realizations taken into account in the average for different lattice sizes .
no abrupt jump that would be the signature of rare events with large contribution is observed .
the estimates of @xmath1 after @xmath104 disorder realizations are compatible within statistical fluctuations for lattice sizes larger than 128 .
our final estimate is @xmath3 . for small lattice sizes
, @xmath1 takes a value close to 10/3 , i.e. the one of the pure 3-state potts model , suggesting that a cross - over between the pure and the random fixed points occurs for @xmath105 .
the data have been analyzed using other fractions of the lattice .
the estimate of @xmath1 tends to increase when going from a large fraction of the lattice to a smaller one : @xmath106 for an interpolation over the right half of the lattice , @xmath107 for an interpolation over the right quarter of the lattice .
a compromise has to be found between a too large fraction for which systematic deviations due to loops may be important and a too small fraction where lattice effects due to the boundaries are important .
we note that , as expected , apllying blindly the interpolation procedure to the whole lattice gives a significantly different estimate @xmath108 . like for the case of the interface length
, we still need to show that a small variation of the disorder amplitude will not induce a large variation of @xmath1 so that @xmath1 would still be compatible with the estimate @xmath109 obtained from the scaling of the interface length .
simulations have been made for other disorder amplitudes @xmath57 but only for lattice sizes @xmath110 and with 3000 disorder realizations .
we have shown in the case of @xmath66 that this lattice size still leads to a small overestimate of @xmath1 .
nevertheless , it should be possible to get a correct order of magnitude of the variation of @xmath1 with @xmath57 . as shown on figure [ fig6 ] , this variation remains small . with a linear interpolation
, one can estimate this variation to be @xmath111 .
although our interface presents loops due to the use of a simple right - turn tie - breaking algorithm , we have shown that they do not affect the scaling dimension of the interface in the pure case . however , systematic deviations between the left - passage probability and the schramm formula were observed .
the parameter @xmath1 can be nevertheless estimated by restricting the interpolation to the right part of the lattice . in the case of the 3-state random bond potts model
, we have measured numerically the sle parameter @xmath1 in two independent ways .
first , we estimated @xmath2 from the fractal dimension @xmath0 of the interface .
this value is compatible , although less accurate , with the estimate obtained in @xcite .
then , the left - passage probability has been fitted with to the schramm formula in the right third of the lattice .
we obtained @xmath3 .
these two compatible values provide further evidences that chordal @xmath4 appropriately describes the geometric properties of interfaces between spin clusters in the 3-state random potts model .
we have not observed the large discrepancy observed in the case of the sos model on a random substrate @xcite .
the institute jean lamour is unit mixte de recherche cnrs number 7198 .
the author would like to express his gratitude to sreedhar dutta for having patiently introduced him to the field of sle .
64 j. cardy ( 1996 ) _ scaling and renormalization in statistical physics _ , cambridge . | we have numerically studied the properties of the interface induced in the ferromagnetic random - bond three - state potts model by symmetry - breaking boundary conditions .
the fractal dimension @xmath0 of the interface was determined .
the corresponding sle parameter @xmath1 was estimated to be @xmath2 , compatible with previous estimate .
on the other hand , we estimated @xmath1 independently from the probability of passage of the interface at the left of a given point .
the numerical data are well reproduced by the schramm theoretical prediction and the fit leads to @xmath3 , in agreement with the first estimate .
this provides evidences that the geometric properties of spin interfaces in the random 3-state potts model may be described by chordal @xmath4 . # |
among active galactic nuclei ( agns ) , bl lac objects are the most extreme class characterized by strong and rapid variability , high polarization , and weak emission lines .
these extreme properties are generally interpreted as a consequence of nonthermal emission from a relativistic jet oriented close to the line of sight ( blandford & k@xmath1nigl 1979 ) . implicit in this hypothesis is that there exist many objects intrinsically identical to bl lac objects , with the relativistic jets oriented at large angle to the line of sight ( in the plane of the sky or perpendicular to our line of sight ) .
while bl lac objects are dominated by relativistically beamed emission from the jets , these objects are most likely dominated by unbeamed , isotropic emission and appear to be very different .
these objects constitute the so - called parent population of bl lac objects , and the unification in this way is usually called the unified scheme " ( antonucci & ulvestad 1985 ) .
it is generally believed that the parent population of bl lac objects is fanaroff - riley class i ( fri , fanaroff & riley 1974 ) radio galaxies ( browne 1983 ; wardle , moore & angel 1984 ) . there is growing evidence for bulk relativistic motion in the jets of blazars ( bl lac objects and ovv quasars ) , and the relativistic beaming model has been widely accepted ( ghisellini et al .
however , almost all arguments for beaming can probably be explained in other ways as well ( urry & padovani 1995 ) .
the strongest challenge for this model remains in testing the unified scheme ( urry & padovani 1990 ) .
numerous studies have found general agreement with the unified scheme of bl lac objects and fri radio galaxies , including studies of 1 ) unbeamed properties of bl lac objects and fri radio galaxies , such as extended radio emission , narrow emission lines , host galaxies and environments , 2 ) luminosity functions of the parent and beamed populations in different bands , 3 ) correlation between core - dominance parameter ( the ratio between core and extended radio flux ) and beamed properties ( for a review see antonucci 1993 ; urry & padovani 1995 ) . recently , hubble space telescope ( @xmath2 ) observations of host galaxies of bl lac objects ( falomo et al .
1997 , urry et al . 1999 ) and of the optical cores of fri radio galaxies ( chiaberge et al . 1999 , 2000 ; capetti & celotti 1999 ) also supported this unified scheme .
furthermore , superluminal motion confirmed in m87 by @xmath2 optical observation ( bireta et al .
1999 ) and in b2 1144 + 35 by vlbi observation ( giovannini et al . 1999 ) provided strong evidence to this unified scheme .
early multiwavelength observations showed that emission from bl lac objects is dominated by synchrotron component with the peak lying in the ir to x - ray energy range . in the 1990 s , with the observations by compton gamma - ray observatory and ground - based cherenkov telescopes
it was found that bl lac objects emit enormous power in rapidly variable high energy gamma rays ( gev and tev ) , indicating a second peak in their overall spectral energy distribution ( sed ; ulrich , maraschi and urry 1997 ; fossati et al .
a practical way to parameterize the different seds is to use the radio - to - x - ray effective spectral index @xmath0 . according to whether @xmath0 is greater than or less than 0.75 ( padovani & giommi 1995 ) bl
lac objects are divided into two subclasses , the low - energy - peaked bl lac objects ( lbls ) which have synchrotron peaks in ir / optical and high - energy - peaked bl lac objects ( hbls ) which have synchrotron peaks in uv / soft x - ray .
most radio - selected bl lac objects ( rbls ) are lbls , while most x - ray - selected bl lac objects ( xbls ) are hbls .
further studies found that the sed difference between lbl and hbl can not be explained in terms of different viewing angles alone , suggesting that there must be some intrinsic differences that cause different seds between lbl and hbl ( sambruna et al .
1996 , 1999 ; georgranopoulos & marscher et al . 1998 ; fossati et al . 1998 ) .
if fri radio galaxies are , as believed today , misaligned bl lac objects , it is expected that the jet - related nonthermal emission from fri radio galaxies also has a double - peak structure in the sed , and that there exist two intrinsically different subclasses of fri radio galaxies : one is lbl - like with @xmath3 , and the other is hbl - like with @xmath4 , corresponding to the parent populations for lbls and hbls , respectively .
therefore , the double - peak characteristics of the sed of the jet - related emission and the existence or nonexistence of two subclasses of fri radio galaxies ( or more strictly speaking , the distribution of @xmath0 in fri radio galaxies ) can be used to test the unified scheme of bl lac objects and fri radio galaxies .
there are some difficulties at present to test the unified scheme in terms of the double - peak characteristics of the sed , because the detectors available are not yet sensitive enough to detect high - energy emission in most of fri radio galaxies . in this paper
we collect x - ray and radio observations available in the literature for most of fris in the zirbel - baum sample ( zirbel & baum 1995 ) , and investigate the distribution of @xmath0 for the sample to test the unified scheme of bl lac objects and fri radio galaxies .
lllrrrrr 0055@xmath501 & 3c29 & 0.0448 & 93.0 & 1 & 11.1 & 2 & 0.82 0104@xmath632 & 3c31 & 0.0167 & 140.0 & 3 & 63.7 & 4 & 0.74 0123@xmath501 & 3c40 & 0.0180 & 100.0 & 1 & 58.5 & 2 & 0.73 0219@xmath642 & 3c66b & 0.0215 & 182.0 & 5 & 141.0 & 4 & 0.71 0255@xmath605 & 3c75 & 0.0232 & 39.0 & 1 & 345.1 & 2 & 0.58 0300@xmath616 & 3c76.1 & 0.0320 & 10.0 & 6 & 33.0 * & 7 & 0.61 0305@xmath603 & 3c78 & 0.0288 & 964.0 & 1 & 698.6 & 2 & 0.72 0314@xmath641 & 3c83.1b & 0.0255 & 40.0 & 8 & 27.0 & 4 & 0.72 0316@xmath641 & 3c84 & 0.0172 & 59600.0 & 9 & 1300.0 & 4 & 0.92 0320@xmath537 & fora & 0.0050 & 26.0 & 1 & 210.1 & 2 & 0.58 0427@xmath553 & pks & 0.0380 & 57.0 & 1 & 116.9 & 2 & 0.66 0430@xmath605 & 3c120 & 0.0330 & 3458.0 & 1 & 9065.5 & 2 & 0.65 0453@xmath520 & b2 & 0.0350 & 40.0 & 1 & 48.7 & 2 & 0.69 0620@xmath552 & pks & 0.0510 & 260.0 & 1 & 488.5 & 2 & 0.66 0625@xmath535 & oh-342 & 0.0550 & 600.0 & 1 & 1327.7 & 2 & 0.65 0722@xmath630 & b2 & 0.0191 & 51.0 & 10 & @xmath7 20.0 & 11 & 0.75 0800@xmath624 & b2 & 0.0433 & 3.0 & 10 & @xmath7 3.0 & 11 & 0.70 0915@xmath511 & pks & 0.0540 & 217.0 & 1 & 7314.8 & 2 & 0.50 1040@xmath631 & b2 & 0.0360 & 55.0 & 10 & 16.0 & 11 & 0.77 1142@xmath619 & 3c264 & 0.0208 & 200.0 & 12 & 486.0 & 4 & 0.65 1144@xmath635 & b2 & 0.0630 & 250.0 & 3 & 170.0 & 11 & 0.72 1216@xmath606 & 3c270 & 0.0060 & 308.0 & 1 & 147.5 & 2 & 0.74 1222@xmath613 & 3c272.1 & 0.0031 & 180.0 & 10 & 26.0 & 4 & 0.81 1228@xmath612 & 3c274 & 0.0043 & 4000.0 & 10 & 1080.0 & 4 & 0.78 1251@xmath512 & 3c278 & 0.0150 & 88.0 & 1 & 79.3 & 2 & 0.71 1254@xmath627 & b2 & 0.0246 & 2.3 & 11 & 25.0 & 11 & 0.57 1256@xmath628 & b2 & 0.0224 & 2.0 & 3 & @xmath7 2.0 & 11 & 0.70 1317@xmath633 & b2 & 0.0379 & 8.0 & 11 & 7.0 & 11 & 0.71 1318@xmath543 & ngc5090 & 0.0110 & 580.0 & 1 & 250.5 & 2 & 0.75 1322@xmath636 & b2 & 0.0175 & 150.0 & 3 & @xmath7 10.0 & 11 & 0.85 1322@xmath542 & cen a & 0.0020 & 6984.0 & 1 & 488.5 & 2 & 0.85 1333@xmath533 & pks & 0.0130 & 297.0 & 1 & 179.5 & 2 & 0.73 1346@xmath626 & b2 & 0.0633 & 53.0 & 10 & 36.0 & 11 & 0.72 1414@xmath611 & 3c296 & 0.0237 & 77.0 & 10 & 57.8 & 4 & 0.72 1422@xmath626 & b2 & 0.0370 & 25.0 & 10 & 4.0 & 11 & 0.80 1511@xmath626 & 3c315 & 0.1080 & 150.0 & 13 & 42.0 * & 7 & 0.74 1514@xmath607 & 3c317 & 0.0342 & 391.0 & 1 & 5167.4 & 2 & 0.55 1553@xmath624 & b2 & 0.0426 & 53.6 & 14 & 11.0 & 11 & 0.79 1610@xmath629 & b2 & 0.0313 & 6.0 & 10 & 12.0 & 11 & 0.66 1621@xmath638 & b2 & 0.0310 & 50.0 & 10 & 16.0 & 11 & 0.76 1626@xmath639 & 3c338 & 0.0298 & 105.0 & 3 & 17.5 & 4 & 0.80 1648@xmath605 & 3c348 & 0.1540 & 10.0 & 1 & 790.5 & 2 & 0.45 1836@xmath617 & 3c386 & 0.0180 & 14.0 & 15 & 30.0 * & 7 & 0.63 1839@xmath548 & pks & 0.1120 & 163.0 & 1 & 229.6 & 2 & 0.68 1855@xmath637 & b2 & 0.0552 & 100.0 & 10 & 35.0 & 11 & 0.76 1954@xmath555 & pks & 0.0600 & 50.0 & 1 & 204.6 & 2 & 0.62 2058@xmath528 & pks & 0.0380 & 63.0 & 1 & 116.9 & 2 & 0.66 2152@xmath569 & pks & 0.0270 & 400.0 & 1 & 903.2 & 2 & 0.65 2212@xmath613 & 3c442a & 0.0263 & 2.0 & 10 & @xmath7 15.0 & 4 & 0.59 2229@xmath639 & 3c449 & 0.0171 & 37.0 & 10 & 16.0 & 4 & 0.75 2335@xmath626 & 3c465 & 0.0293 & 270.0 & 3 & 65.8 & 4 & 0.78
our sample , as listed in table 1 , comprises most of fri radio galaxies in the sample of radio - loud elliptical galaxies compiled by zirbel & baum ( 1995 ) .
the zirbel - baum sample is the largest known sample of radio - loud elliptical galaxies ( excluding quasars ) .
the radio luminosity ranges from the radio - quiet " ellipticals of phillips et al .
( 1986 ) to the more powerful radio galaxies of mccarthy ( 1988 ) , nicely covering fris and friis which are located within the transition region in radio power ( zirbel & baum 1995 ) . for an fri radio galaxy ,
zirbel & baum ( 1995 ) defined it solely on the basis of radio morphology as a radio galaxy in which the radio emission peaks near the center of the galaxy and the twin jets fade with distance from the center producing the diffuse , plumelike and edge - dimmed radio lobes .
complete samples of fri radio galaxies , such as 3cr ( laing et al 1983 ) , b2 ( ulrich 1989 ) , wall - peacock ( wall & peacock 1985 ) , and other samples are all included in the zirbel - baum sample .
it is therefore thought that although the zirbel - baum fri sample itself is not complete , it can represent the fri population ( e.g. kollgaard et al .
1996 ; urry & padovani 1995 ) . some sources have been reclassified .
for example , 3c 28 , 3c 293 , 3c 305 , 3c 310 , 3c 346 , 3c 433 and ngc 6109 which were identified to be fris by laing et al .
( 1983 ) have been reclassified as friis , 3c 40 which was identified to be an frii by morganti et al .
( 1993 ) has been reclassified as an fri , and caganoff s sources ( caganoff 1988 ) have been all reclassified from the original radio maps ( zirbel & baum 1995 ) .
although it is true that some of frii radio galaxies may also be parent population of bl lac objects which are probably identical to lbls , we exclude possible friis which have been identified as fris initially but have been reclassified as friis later , since our purpose is to investigate whether there exist hbl - like and lbl - like subclasses of fri radio galaxy .
for some sources , x - ray data are not available at present .
these sources are therefore excluded from our sample , yet our sample comprises over 70% fri radio galaxies in the zirbel - baum sample . in table 1 ,
the columns are : ( 1 ) iau name ; ( 2 ) other name ; ( 3 ) redshift ; ( 4 ) flux density of the core at 5ghz in unit of mjy ; ( 5 ) reference for the radio flux density ; ( 6 ) flux density of the central unresolved component at 1kev in unit of njy ; ( 7 ) reference for x - ray flux ; ( 8) effective spectral indices between 5ghz and 1kev corrected for beaming . for some objects ,
x - ray flux densities are calculated from broad band flux with the photon index ( or spectral index ) given in the corresponding reference .
effective spectral index @xmath0 between 5ghz and 1kev is calculated as ( lamer et al .
1994 ) @xmath8 and between 5ghz and 2kev as ( perlman et al .
1996 ) @xmath9 where @xmath10 , @xmath11 and @xmath12 are flux densities at 5ghz , 1kev and 2kev in the rest frame of the source , respectively .
the flux densities in table 1 have been all @xmath13-corrected before computing @xmath0 .
since the radio cores of fris have a flat spectrum , we take @xmath14=0.0 for the @xmath13-correction when there is no @xmath14 in the literature , and @xmath15=0.8 , the typical value for fri class .
the value of @xmath0 in table 1 has been corrected for beaming as well . according to the unified scheme
, an fri radio galaxy with @xmath16 is viewed at a large angle to the line of sight . when viewed at a small angle , it will appear as a bl lac object with @xmath17 .
the transformation law for the change of @xmath0 due to relativistic beaming is given by chiaberge et al .
( 2000 ) as @xmath18 where @xmath19 and @xmath20 are doppler factors for bl lac objects and fri radio galaxies , respectively .
the doppler factor is defined as @xmath21^{-1}$ ] , where @xmath22 is the speed of emitting plasma in unit of the speed of light , @xmath23 is lorentz factor @xmath24 and @xmath25 is the angle to the line of sight .
we take typical value for viewing angle and lorentz factor .
the typical viewing angle of fri radio galaxies is @xmath26 , and for bl lac objects , the typical viewing angle is @xmath27 ( e.g. padovani 1999 ) .
the typical lorentz factor for relativistic jets is @xmath28 ( padovani 1999 ; param et al .
1996 ; marscher 1993 ) .
as mentioned in the introduction section , a practical way to parameterize the different seds is to use the radio - to - x - ray spectral index , @xmath0 , and according to whether @xmath0 is greater than or less than 0.75 ( padovani & giommi 1995 ) bl lac objects are divided into two subclasses , lbl and hbl .
figure 1a displays the distribution of @xmath0 for our sample of 51 fri radio galaxies . for comparison ,
figure 1b illustrates the distribution of @xmath0 for bl lac objects in the samples of emss ( the most x - ray - dominated bl lac objects ) , @xmath29-green bank ( rgb , intermediate ) and 1jy ( the most radio - dominated bl lac objects , see the fourth paragraph of this section for detailed properties of bl lac samples ) . for rgb objects ,
@xmath0 is calculated with the data in laurent - muehleisen et al .
( 1999 ) , and for 1jy and emss objects , @xmath0 is taken from sambruna et al .
we adopt the data in the rgb sample for overlapping sources between the rgb and emss or 1jy sample .
important features are seen in figure 1a and table 1 : a ) the range of @xmath0 of fri radio galaxies is 0.45 to 0.92 , which is almost the same as that of bl lac objects ( 0.45 to 1.01 , see figure 1b ) ; b ) there is a peak at @xmath300.70 and the average of @xmath0 for the whole sample is 0.70 .
( the average of @xmath0 of bl lac objects in fig .
1b is 0.69 . ) ; c ) 38 out of total 51 fris have @xmath4 .
these suggest that hbl - like fri radio galaxies are not rare exceptions as in the case for flat spectral radio quasars ( fsrqs ; padovani , giommi , & fiore 1997 ) , but statistically constitute a subclass which populates about half of the sample .
these results shows that there exist two subclasses of fri radio galaxies : one is hbl - like and the other is lbl - like .
investigations of multifrequency spectral energy distribution from the radio through the x - ray bands for bl lac objects and fsrqs show that the different seds between hbl and lbl can not be explained in terms of different viewing angles alone ( sambruna et al .
1996 ; fossati et al . 1998 ) .
there must be some intrinsic differences that cause different seds between lbl and hbl .
as their parent population , fri radio galaxies should exhibit similar differences , but should not be what was once believed that every fri could appears as an hbl or an lbl if viewed at different angles to the line of sight .
therefore , the existence of the hbl - like and lbl - like subclasses of fri radio galaxies supports the unified scheme of bl lac objects and fri radio galaxies .
in fact , as early as in 1992 , based on vla observation for complete samples of lbl and hbl , kollgaard et al .
( 1992 ) and laurent - muehleisen et al .
( 1993 ) found that only those fri radio galaxies with intrinsically stronger cores would appear as an lbl if viewed at a small angle to the jet direction , while hbl may be found among the weaker cored fri radio galaxies .
this conflicted with the prevailing point of view at that time that xbls ( mainly hbls ) were intrinsically the same as rbls ( mainly lbls ) , but now has actually become another evidence of the unified scheme , and is consistent with our results .
in addition , the result of kollgaard et al . also suggests that care should be taken when using flux - limited complete samples to test the unified scheme , since there may be few hbl - like fris in the samples of brighter radio source . in fig .
1a , there is only one peak at @xmath31 in the distribution of @xmath0 . it had been known for long that the distribution of bl lac objects is bimodal which peaks at @xmath32 and 0.9 ( e.g. stocke et al .
1985 , 1991 ; padovani & giommi 1995 , 1996 ; sambruna et al . 1996 ; fossati et al . 1998 ) .
however , a large fraction of newly discovered bl lac objects have been found to fill in the gap between the two peaks , which are called intermediate bl lac objects ( ibls , perlman et al . 1998 ; caccianiga et al .
1999 ; laurent - muehleisen 1997 ; laurent - muehleisen et al .
1999 ) . probably , as laurent - muehleisen et al .
( 1999 ) pointed out , the bimodal distribution of bl lac objects found in earlier studies was caused by observational selection effects , and the true distribution of x - ray - to - radio flux ratio @xmath33 ( equals the distribution of @xmath0 ) is unimodal ( laurent - muehleisen et al .
a kolmogorov - smirnov test shows that the distributions of @xmath0 for fri radio galaxies and bl lac objects in figs .
1a and 1b are indistinguishable at the level of 90 percent . however , further quantitative test is necessary in the future , because it seems that none of the flux - limited complete samples of bl lac objects available at present can represent the whole class of bl lac objects .
recently , chiaberge et al . ( 2000 ) tested the unification of bl lac objects and fri radio galaxies by comparing the core emission of radio galaxies with those of bl lac objects , taking advantage of the @xmath2 optical observation for the core in 3cr fri radio galaxies . from the comparison of the optical and radio emission of fris and bl lac objects , they inferred lorentz factors of @xmath34 , which are typical for a relativistic jet ( padovani 1999 ; param et al .
1996 ; marscher 1993 ) , supporting the unified scheme .
their study also shows that in the @xmath35 ( optical and radio luminosities of the core ) plane , hbl and lbl move on different debeaming trails , and that debeaming trail of lbl crosses the 3cr fri region ( both the simple emission component model and two - velocity jet model ) . by taking the typical value for the lorentz factor or assuming a two - velocity structures for the jet
, the location of debeamed bl lac objects in the 1jy sample in the @xmath35 plane coincides with that of fris in the 3cr sample , indicating that fris in the 3cr sample are lbl - like or ibl - like .
this is also consistent with our result .
it is notable that the percentage of lbl - like fris in our sample is less than 50% after beaming correction .
this may indicate that some friis or at least the intermediate sources ( fri / ii ) should be included in the parent population of bl lac objects , which was also suggested by vla observations ( kollgaard et al .
1992 ) and optical spectroscopic observations ( owen et al . 1996 ) .
another possible explanation is that this may be caused by the limited angular resolution of x - ray observations .
for example , the angular resolution of @xmath29 with the hri is only about 5@xmath36 .
the x - ray flux of the central unresolved component of an fri radio galaxy is probably not contributed by the vla radio core alone . in the near future ,
chandra x - ray observatory ( cxo ) will observe the central region of fri radio galaxies with much higher spatial resolution and sensitivity , obtaining a much better and larger x - ray - observed sample to test the unified scheme more reliably .
in addition , with the cxo s high sensitive and high spectral resolution observation , the existence of hbl - like and lbl - like fri radio galaxies can be tested directly by comparing the x - ray spectra .
since the hbl - like fris peak the synchrotron emission at uv / x - ray energy range , the x - rays are an extension of synchrotron emission with a steeper spectrum , and the shape of the optical - to - x - ray continuum is convex ( @xmath37 ) ; while for most lbl - like fris , the x - rays are dominated by inverse - compton emission with a flatter spectrum , and the shape of the optical - to - x - ray continuum is concave ( @xmath38 ) .
therefore , hbl - like fris can be distinguished from lbl - like fris in the optical to soft x - ray energy range . in summary ,
the range of @xmath0 for fri radio galaxies is almost the same as that for bl lac objects , the distribution of @xmath0 probably peaks at the same position as bl lac objects , and the distribution of @xmath0 for fris is similar to that for bl lac objects .
these suggest that there exist two subclasses of fri radio galaxies : one is hbl - like , and the other is lbl - like , corresponding to hbls and lbls , respectively .
this result supports the unified scheme of bl lac objects and fri radio galaxies .
antonucci , r.r.j .
1993 , ara&a , 31 , 473 antonucci , r.r.j . &
ulvestad , j. s. 1985 , apj , 294 , 158 bireta , j.a . , sparks , w.b . ,
& macchetto f. 1999 , apj , 520 , 621 blandford , r. d. , & k@xmath1nigl , a. 1979 , apj , 232 , 34 browne , i.w.a .
1983 , mnras , 204 , 23p caccianiga , a. , et al . 1999 ,
apj , 513 , 51 caganoff , s. 1988 , ph.d .
thesis , the australian national university canosa , c.m . , worrall , d.m . ,
hardcastle , m.j . ,
birkinshaw , m. 1999 , mnras , 310 , 30 capetti , a. & celotti , a. 1999 , mnras , 304 , 434 chiaberge , m. , celotti , a. , capetti , a. , & ghisellini , g. 2000 , a&a , 358 , 104 chiaberge , m. , celotti , a. , capetti , a. 1999 , a&a , 349 , 77 fabbiano , g. , trinchieri , g. , elvis , m. , et al .
1984 , apj , 277 , 115 falomo , r. et al .
1997 , apj , 476 , 113 fanaroff , b. , & riley , j.m . 1974 ,
mnras , 167 , 31p fanti , c , fanti , r , de ruiter h.r . , parma , p. 1987
, a&as , 69 , 57 fossati , g. , maraschi , l. , celotti , a. , et al . 1998 ,
mnras , 299,433 gavazzi , g. , perola , g.c . ,
& jaffe , w. 1981 , a&a , 103 , 35 georgranopoulos , m. , & marscher , a.p .
1998 , apj , 506 , 621 ghisellini , g. , padovani , p. , celotti , a. , & maraschi , l. 1993 , apj , 407 , 65 giovannini , g. , et al . 1999 ,
apj , 522 , 101 giovannini , g. , feretti , l. , gregorini , l. , parma , p. 1988
, a&a , 199 , 73 giovannini , g. , feretti , l. , comoretto , g. 1990 , apj , 358 , 159 hardcastle , m.j . , & worrall , d.m .
1999 , mnras , 309 , 969 h@xmath1gbom , j.a . 1979 , a&as , 36 , 173 kollgaard , r.i . , et al . 1996 , apj , 465 , 115 kollgaard , r.i . ,
wardle , j.f.c . ,
roberts , d.h . , & gabuzda , d.c .
1992 , aj , 104 , 1687 laing , r.a . ,
riley , j.m .
, & longair , m.s .
1983 , mnras , 204 , 151 lamer , g. , brunner , h. , & staubert , r. 1996 , a&a , 311 , 384 laurent - muehleisen , s.a . , kollgaard , r.i . ,
feigelson , e.d . , et al .
1999 , apj , 525 , 127 laurent - muehleisen , s.a .
1997 , pasp , 109 , 341 laurent - muehleisen , s.a . ,
kollgaard , r.i . ,
moellenbrock , g.a .
, & feigelson , e.d .
1993 , aj , 106 , 875 leahy , j.p . ,
j@xmath39gers , w.j . ,
& pooley , g.g .
1986 , a&a , 156 , 234 macklin , j.t . 1983 , mnras , 203 , 147 marscher , a.p .
1993 , in astrophysical jets , ed .
d. burgarella , m. livio , & c.p .
odea ( cambridge : cambridge university press ) , 73 mccarthy , p.j .
1988 , ph.d .
thesis , university of california at berkeley morganti , r. , killeen , n.e.b .
, tadhunter , c.n .
1993 , mnras , 263 , 1023 noordam , j.e .
, & de bruyn , a.g . 1982 .
nat , 299 , 597 odea , c.p .
, & owen , f.n .
1985 , aj , 90 , 927 owen , p. n. , ledlow , m.j . , &
keel , w.c .
1996 , aj , 111 , 53 padovani , p. 1999
series , 159 , 339 padovani , p. and giommi , p. , & fiore , f. 1997 , mnras , 284 , 569 padovani , p. and giommi , p. 1996 , mnras , 279 , 526 padovani , p. and giommi , p. 1995 , apj , 444 , 567 param , p. , de ruiter , h.r . , & fanti , r. 1996 , in iau symp . 175 , extragalactic radio sources , ed .
r. ekers , c.fanti , & l.padrielli ( dordrecht : reidel ) , 137 perlman , e.s .
, padovani , p. , giommi , p. , et al .
1998 , aj , 115 , 1253 perlman , e.s .
, et al . 1996 ,
apjs , 104 , 251 phillips , m.m . , et al .
1986 , aj , 91 , 1061 sambruna , r.m . , maraschi , l. , & urry , c.m .
1996 , apj , 463 , 444 sambruna , r.m . ,
et al . 1999 ,
apj , 515 , 140 siebert , j. , brinkmann , w. , morganti , r. , et al .
1996 , mnras , 279 , 1331 stocke , j.t . ,
1991 , apjs , 76 , 813 stocke , j.t . , et al . 1985 ,
apj , 298 , 619 strom , r.g . , willis , a.g . ,
wilson , a.s .
1978 , a&a , 68 , 367 ulrich , m .- h . , maraschi , l. , & urry , c.m .
1997 , ara&a , 35 , 445 ulrich , m .- h .
1989 , in bl lac objects , edited by l. maraschi , t. maccacaro and m .- h .
ulrich ( springer , berlin ) , 45 urry , c.m .
, et al . 1999 ,
apj , 512 , 88 urry , c.m . , padovani , p. 1995 ,
pasp , 107 , 804 urry , c.m .
, padovani , p. 1990 , apj , 356 , 75 wardle , j.f.c . , moore , r.l .
, & angel , j.r.p .
1984 , apj , 279 , 93 wall , j.v . ,
peacock , j.a .
, 1985 , mnras , 216 , 173 zirbel , e.l . , & baum , s.a .
1995 , apj , 448 , 548 | in this paper , we collect radio and x - ray observations for most fanaroff - riley i ( fri ) radio galaxies in the zirbel - baum radio galaxy sample , and investigate the distribution of the radio - to - x - ray effective spectral index , @xmath0 , to test the unified scheme of bl lac objects and fri radio galaxies .
it is found that the range of @xmath0 for fri radio galaxies is almost the same as that for bl lac objects , that the distribution of @xmath0 probably peaks at the same position as bl lac objects , and that the distribution of @xmath0 for fris is similar to that for bl lac objects .
these suggest that there exist two subclasses of fri radio galaxies : one is hbl - like , and the other is lbl - like , corresponding to high - energy - peaked ( hbl ) and low energy - peaked ( lbl ) bl lac objects , respectively .
this result is consistent with previous vla observations , and supports the unified scheme of bl lac objects and fri radio galaxies . |
we try to find changes in the properties of grbs at different redshifts which could be related to source evolution .
we consider all the 149 events detected by swift @xcite , @xcite between january 26 , 2005 and july 15 , 2009 for which the redshift has been at least tentatively measured .
we use the table given at the heasarc website .
figure 1 shows the histogram of all grb redshifts detected until july 15 , 2009 .
data for figure 1 were taken from the above quoted website and from gcn .
we also show the histogram for events detected by hete-2 and _ beppo_sax , although the trigger criteria were not the same for all three experiments .
the swift table gives also the bat fluence ( 15 - 150 kev ) and the bat t@xmath1 ( 15 - 350 kev ) .
we consider both the values in the observer s frame and the ones converted in the restframe .
the redshift in the events goes from 0.0331 to 8.26 .
note that the lowest value of the redshift until now is 0.0085 for grb980425 , detected by _
beppo_sax .
the scatter diagrams of those quantities , namely bat t@xmath1 and fluence , are shown in figures 2 and 4 respectively , both in the observer s and in the restframe . for lack of some data only a total of 139 events could actually be used . for comparison
we also show t@xmath1 versus redshift ( figure 3 ) and fluence versus redshift ( figure 5 ) as obtained using the gamma ray burst monitor ( grbm ) @xcite , on board of _
beppo_sax ( energy bands : @xmath2 kev for t@xmath1 and 40700 kev for fluence ) and french gamma - ray telescope ( fregate ) @xcite , on board of hete-2 ( energy bands : 680 kev for t@xmath1 and 230 kev for fluence ) . as noted by dr .
upendra desai @xcite in figure 2 and even more in figure 4 there seems to be a lack of
low t@xmath1low fluence " events at @xmath3 and @xmath4 , but unfortunately the number of grbs is still too small to allow us to confirm this possibility .
we recall that , for example , in the case of quasars the periodicity of redshifts is a problem which has been debated for many years @xcite . as shown in figure [ sw_z - ttrtrmed - fig ] , we also tried to compensate for the fact that the values , taken at the same energy range in the observer s frame , originate from different energy ranges in the event s restframe .
t@xmath1 in the restframe is the observed value divided by @xmath5 and the fluence in the restframe is the observed value multiplied by @xmath5 , but their respective energy intervals in the restframe are also multiplied by @xmath5 . by using the fenimore @xcite correlation between peak duration and energy , and assuming that it can be applied also to t@xmath1 , we try to take into account that , for long bursts , the duration normally decreases with energy . in figure 7
we show the scatter diagram of fluence versus t@xmath1 , again in the observer s and restframe for the swift events , color coded for six redshift intervals .
we also computed the fluence
t@xmath1 correlation coefficients for the same redshift groups shown in figure 6 ( see table [ coeff - tab ] ) .
the correlation for all redshifts together is @xmath6 , both for observed and restframe values , non weighted , while it is @xmath7 and @xmath8 for observed and restframe weighted values . [
cols="^,^,^,^,^,^,^",options="header " , ] [ coeff - tab ]
we considered 139 grbs at different redshifts , all of them detected by the same experiment , swift - bat , hoping to find proof of evolution with @xmath9 . except for
the well known two groups of short " and long " grbs , which appear to be a little less well defined in the restframe , no such proof is evident .
we can see from the plots ( figures 2 and 4 ) that the number of events obviously becomes smaller with redshift , but both t@xmath1 and fluence hardly change in their average log value , where we find them also for the events at the largest redshifts .
thus we conclude that the probability of having grb with those values is higher , even at large @xmath9 .
we conclude that no redshift selection or evolution can yet be inferred from our plots .
even grb090423 , the one detected at the largest @xmath9 until now , lies just in the middle of the distribution in the observer s restframe .
not surprisingly , it is evident from the restframe plots ( figure 7 , right panel ) that fluence increases with t@xmath1 practically for all redshifts .
bursts at large redshifts have higher fluences , but we must remember that they originate from higher energy ranges and that detection thresholds favor distant intense events . 9 salvaterra , r. , et al .
2009 , nature , 461 , 1258 tanvir , n.r . ,
2009 , nature , 461 , 1254 barthelmy , s.d . ,
2005 , ssr , 120 , 143 gehrels , n. , et al , 2007 , njph , 9 , 37 frontera , f. et al .
2009 , apjs , 180 , 192 pelangeon , a. et al .
2008 , a&a , 491 , 157 desai , u.d .
private communication lopez - corredoira , m. 2009 , arxiv0910.4297 fenimore , e.e . ,
1995 , apj , 448 , l101 | grbs are now detected up to @xmath0 @xcite,@xcite .
we try to find differences , in their restframe properties , which could be related either to distance or to observing conditions . |
in a @xmath0sr experiment the muon - spin precession frequency @xmath1 can be used to deduce the local magnetic field @xmath5 at the muon site .
if there are a number of muon sites with different local fields @xmath6 , then the @xmath0sr signal can contain a number of components with frequecies @xmath7 .
in complex systems it can be highly non - trivial to determine the precise location of the muon site or sites . however , it is nevertheless useful to attempt to extract an estimate of the magnitude of the moment of the magnetic species producing the local field , even in the face of ignorance of the location of the muon site or sites . in this paper
we describe a method to attempt this using bayesian inference .
the paper is structured as follows : in sections 2 and 3 we review the theory of dipolar fields and bayes theorem respectively to provide the necessary background to the calculation which is described in section 4 .
an implanted muon spin precesses around a local magnetic field , @xmath8 , with a frequency @xmath9 , where @xmath10 . the magnetic field
@xmath8 at the muon site is given by @xmath11 where @xmath12 represents the applied field ( zero in the experiments considered here ) , @xmath13 is the dipolar field from magnetic ions , @xmath14 is the lorentz field , @xmath15 is the demagnetizing field from the sample surface and @xmath16 is the contact hyperfine field caused by any spin density overlapping with the muon wavefunction . in antiferromagnets
the lorentz and demagnetizing fields vanish ( in polycrystalline ferromagnets they cancel to some extent ) .
the contact hyperfine field is hard to estimate but we will neglect it .
the remaining term is the dipolar field @xmath17 and is a function of the muon - site @xmath18 .
it can be written as @xmath19 a sum over the magnetic ions ; the magnetic moment of the @xmath20th ion is @xmath21 . in eq .
( [ eq : localfield ] ) , @xmath22 is the dipolar tensor given by @xmath23 where @xmath24 and @xmath25 is the kronecker delta ( @xmath26 if @xmath27 , else @xmath28 ) .
the behaviour of this tensor is dominated by the arrangement of the nearest - neighbour magnetic ions and leads to a non - zero local magnetic field for almost all possible muon sites , even in an antiferromagnetically ordered system @xcite .
the sum in eq .
( [ eq : localfield ] ) is taken over the infinite lattice , but it is well known @xcite that this sum converges in such a way that it is necessary only to sum over points inside a sphere centred on @xmath18 with sufficiently large radius .
an alternative method of calculation is provided by the method of ewald summation ( for details see @xcite ) .
we recap some elementary probability theory @xcite . the _ conditional probability _ @xmath29 is the probability that event @xmath30 occurs _ given _ that event @xmath31 has happened . the _ joint probability _
@xmath32 is the probability that event @xmath30 and event @xmath31 both occur .
the joint probability @xmath32 is equal to the probability that event b occurred multiplied by the probability that a occurred , given that b did , i.e. , @xmath33 and , equally well , @xmath34 now consider the case where there are a number of mutually exclusive events @xmath35 such that @xmath36 then we can write the probability of some other event x as @xmath37 in very general terms , one can say that given some hypothesis h there usually exists some computational strategy to evaluate the probability of a particular outcome o assuming that hypothesis to be correct ( i.e. , there is some method to compute the quantity @xmath38 ) .
however , what you often want to do is the reverse of this : you know the outcome because it has actually occurred and you want to choose an explanation out of the possible hypotheses . in other words , given the outcome you want to know the probability that the hypothesis is true , and the problem is that @xmath39 is typically much more challenging to evaluate .
the needed transformation of @xmath38 into @xmath39 can be accomplished using _ bayes theorem _ ( named after thomas bayes ( 17021761 ) , although the modern form is due to laplace ) .
this theorem can be stated as follows : @xmath40 here @xmath41 is called the _ prior probability _ , since it is the probability of @xmath30 occurring without any knowledge as to the outcome of @xmath31 .
the quantity which you derive is @xmath42 , the _
posterior probability_. the proof of bayes theorem is very simple : one simply equates eqs .
( [ eq : acapb1 ] ) and ( [ eq : acapb2 ] ) and rearranges . for the purpose of this paper , we will write bayes theorem using eq .
( [ bayessum ] ) as @xmath43
since the positive muon seeks out areas of negative charge density , constraints can be placed on the likely location of stopped muons . for example , the muon is unlikely to stop close to the positively charged ions in a system . in many oxides ,
muons have been shown to stop around @xmath44 from an o@xmath45 ion @xcite . in our calculations ,
we assume a magnetic moment @xmath0 on the magnetic species in our material and consider a particular magnetic structure .
positions in the unit cell are then generated at random and , provided the relevant constraints are satisfied , the dipole field is calculated at each of them .
after many such randomly generated positions , one obtains a distribution of dipole fields .
( this distribution sometimes have sharp features associated with them which are van hove singularities @xcite . ) the magnitudes of the resulting dipole fields are then converted into muon precession frequencies , and the resulting histogram yields the probability density function ( pdf ) @xmath4 , evaluated as a function of precession frequency @xmath1 divided by magnetic moment @xmath0 ( since the precession frequency scales with the magnetic moment ) .
this function @xmath4 allows us to evaluate @xmath46 of eq .
( [ eq : bayes ] ) . we can write @xmath47 the function @xmath4 is normalized so that @xmath48 , and hence the factor of @xmath49 is needed in eq .
( [ eq : bayesexpression ] ) so that @xmath50 .
an example of this approach is shown in fig .
1(a ) for a cubic lattice of antiferromagnetically coupled spins .
the solid line shows the case when a simple constraint is applied so that the muon is not permitted to stop at a site closer than a particular critical distance to the magnetic moments , resulting in a cut - off of the tail at high frequency .
( with antiferromagnetic wave vector @xmath51 ) .
the lattice parameter is @xmath52 and the size of the moment is @xmath0 @xcite .
the solid ( dashed ) line shows the distribution with a cut - off ( not ) applied .
( b ) the extracted pdf for the moment given a particular observed frequency ( @xmath53 ) . ] since @xmath1 is obtained from a real experiment , what we would like to know is @xmath3 , the pdf of @xmath0 _ given _ the observed @xmath1 .
this can be obtained from our calculated @xmath4 using bayes theorem in the form of eq .
( [ eq : bayes ] ) , which yields @xmath54 where we have assumed a prior probability [ @xmath55 for the magnetic moment that is uniform between zero and @xmath56 , and so @xmath57 is replaced by the uniform probability density @xmath58 [ which cancels on the top and bottom of eq .
( [ eq : final ] ) ] .
we choose @xmath56 to take a large value , although we have found that our results are insensitive to the precise value of @xmath59 .
a very simple example of this approach is shown in fig .
when multiple frequencies @xmath60 are present in the spectra , it is necessary to multiply their probabilities of observation in order to obtain the chance of their simultaneous observation , so we evaluate @xmath61 , where @xmath62 is the error on the fitted frequency .
we have now applied this technique to @xmath0sr data a variety of real systems in which the muon site is not known .
these include ba@xmath63naoso@xmath64 in which we can show from the observed precession frequencies that the magnetic ground state is most likely to be low - moment ( @xmath65@xmath66 ) ferromagnetism and not canted antiferromagnetism @xcite .
we have also used it to show a reduced moment in the two - dimensional molecular magnet [ cu(hf@xmath63)(pyz)@xmath63]bf@xmath68 @xcite and in the pnictide superconductor nafeas @xcite . in all these cases
we do not have a priori information concerning the muon site but can nevertheless place bounds upon the magnetic moment from the observed precession signal using this technique .
a possible drawback that should be borne in mind is that the hyperfine contribution to the local field is neglected and if this is significant it could affect the conclusions drawn .
as many of the systems examined so far using this technique have localized , reduced moments and lower - frequency precession signals , it is probable that the hyperfine contribution is not significant in these cases .
we thank epsrc ( uk ) for financial support .
00 s. j. blundell , phil .
trans . r. soc .
a357 ( 1999 ) 2923 .
a. j. steele , t. lancaster , s. j. blundell , p. j. baker , f. l. pratt , c. baines , m. m. conner , h. i. southerland , j. l. manson , j. a. schlueter , phys .
b 84 ( 2011 ) 064412
. j. d. wright , t. lancaster , i. franke , a. j. steele , j. s. mller , m. j. pitcher , a. j. corkett , d. r. parker , s. j. clarke , f. l. pratt , p. j. baker , s. j. blundell , submitted . | a significant challenge in zero - field @xmath0sr experiments arises from the uncertainty in the muon site .
it is possible to calculate the dipole field ( and hence precession frequency @xmath1 ) at any particular site given the magnetic moment @xmath0 and magnetic structure .
one can also evaluate @xmath2 , the probability distribution function of @xmath1 assuming that the muon site can be anywhere within the unit cell with equal probability , excluding physically forbidden sites . since @xmath1 is obtained from experiment , what we would like to know is @xmath3 , the probability density function of @xmath0 _ given _ the observed @xmath1 .
this can be obtained from our calculated @xmath4 using bayes theorem .
we describe an approach to this problem which we have used to extract information about real systems including a low - moment osmate compound , a family of molecular magnets , and an iron - arsenide compound .
muon - spin rotation , dipole fields , muon sites |
dwarf galaxies ( m@xmath14 ) are the most abundant class of galaxies in the universe .
they occupy an important part of parameter space for understanding the feedback processes that seem to govern the relationships between dark halo mass , baryon fraction , and star formation efficiency .
furthermore , their progenitors at high redshift may have played an important role in reionizing the universe @xcite .
despite the ubiquity of dwarf galaxies , it is challenging to reliably measure their physical properties and put them in their appropriate cosmological context ( e.g. , * ? ? ? * ) .
most dwarf galaxies are undetectable beyond redshift @xmath15 @xmath212 , even in the hubble ultra deep field or with planned jwst observations @xcite .
thus , observations of local group galaxies have set the benchmark for the accuracy and precision with which ancient star formation rates and chemical evolution histories can be measured ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) .
long - lived main - sequence stars born at lookback times @xmath1610 gyr , corresponding to @xmath15 @xmath17 2 , are a direct probe of galaxy evolution in the universe from the earliest star - forming period , through the epoch of reionization and its aftermath . in this paper
we present a photometric analysis of the understudied star cluster at the center of ddo 216 ( the pegasus dwarf irregular , pegdig , ugc 12613 ) , which we observed serendipitously during our hst program to measure the complete star formation history of this galaxy .
pegdig ( m@xmath18 = @xmath012.5 @xmath10.2 ) is roughly a magnitude fainter than the fornax and sagittarius dwarfs , which makes it one of the least luminous galaxies known to host a cluster near the peak of the globular cluster luminosity function ( see * ? ? ?
* ; * ? ? ?
* for examples of similar clusters in dwarfs with m@xmath18 @xmath1911.5 ) .
first , we review the basic parameters of the galaxy , and then describe our observations and reductions in section [ sec : obs ] .
our analysis of the structure and content of the star cluster ddo216-a1 , including age and metallicity estimates based on the color - magnitude diagram and an analysis of the variable star population of the cluster , is given in section [ sec : clus ] .
we place ddo216-a1 in context with the population of massive star clusters in dwarf galaxies and summarise our results in section [ sec : summ ] .
given deep enough observations , it is relatively straightforward to derive star formation rates at high lookback times for galaxies within the local group . it is more difficult to identify the triggers of star formation .
for example , the majority of dwarf - dwarf major mergers are expected to have occured in the first few billion years after galaxy formation began ( e.g. , * ? ? ? * ) , but because of the destructive nature of mergers most of the obvious evidence for merger activity will have long since vanished .
globular clusters are an important window into this process because they require extreme conditions to form , suggestive of vigorous star formation in the mode often associated with galaxy mergers and interactions ( e.g. , * ? ? ?
* and references therein ) .
globular clusters are tightly bound and will typically survive for a hubble time unless disrupted in a hostile tidal environment , but the relatively shallow potential wells of dwarf galaxies are not generally conducive to cluster disruption .
the globular cluster populations of dwarf galaxies are thus an object of strong interest for the study of the early evolution of these systems and for attempts to gauge the importance of various modes of star formation in small galaxies . with the ability of hst imaging to differentiate between star clusters , background galaxies , and density variations in field populations ,
there has been renewed attention to the study of globular clusters in extragalactic systems in recent decades . as a result , many globular and open clusters are known in dwarf galaxies in the local group and beyond , in enough numbers to make statistical associations between properties like host galaxy morphology and cluster colours and sizes ( e.g. , * ? ? ?
* ; * ? ? ?
these span a range of sizes from extremely luminous and dense nuclear star clusters to low - mass open cluster or association analogues , in galaxies down to some of the least luminous known . in the local group , recent
work has discovered examples of modest star clusters even among the smallest galaxies ( e.g. , * ? ? ?
* ) , and more massive , sometimes extended clusters in some of the larger irregular galaxies @xcite . in light of these discoveries and others , @xcite have suggested that some of the outer halo globular clusters thought to be galactic globular clusters may in fact be hosted by undiscovered low surface brightness galaxies .
however , the lowest - luminosity local group galaxies with catalogued globular clusters similar to the massive and dense globulars of the milky way are fornax and sagittarius ( m@xmath18 = @xmath013.4 and @xmath2013.5 , respectively ) .
the pegasus dwarf irregular galaxy , pegdig , was discovered by a.g .
wilson in the early 1950 s on palomar schmidt plates @xcite . from early on it was considered to be a candidate member of the local group with a distance of @xmath111 mpc .
the case for membership was supported by the negative heliocentric radial velocity found by @xcite .
pegdig is considered a distant m31 satellite ( d@xmath21 @xmath2470 kpc , * ? ? ?
* ) ; it is not proven that it has ever interacted with m31 , although it is more likely than not that pegdig has previously been within m31 s virial radius @xcite .
pegdig is fairly isolated at the present time , its nearest neighbor being the m31 satellite and vi , just over 200 kpc away .
it is quite unlikely that pegdig has had strong tidal interactions with any other known galaxy during the past several gyr .
pegdig is a fairly typical small irregular galaxy , with roughly 1:1 gas to stellar mass ratio , although it has virtually no current star formation as measured by h@xmath22 emission @xcite .
this leads to its classification as a transition type dwarf , with properties intermediate between the spheroidal ( dsph ) and irregular ( dirr ) types @xcite .
it has an ordinary metallicity of [ fe / h ] @xmath2 @xmath01.4 @xmath10.3 for its stellar mass of @xmath210@xmath23 m@xmath13 @xcite . unlike the spheroidal galaxies
, it appears to be rotating ; both hi @xcite and stellar @xcite data suggest a rotation speed ( not corrected for inclination ) of @xmath21520 km / s .
@xcite drew attention to the cometary appearance of the neutral gas , and attributed the asymmetric morphology to ram pressure stripping by diffuse gas in the local group , although this conclusion is disputed , based on much deeper hi observations , by @xcite .
the star formation history of pegdig has been estimated from ground - based data reaching a limiting absolute magnitude of m@xmath24 @xmath2 @xmath02.5 @xcite , and from hst / wfpc2 observations @xcite reaching @xmath22.5 mag deeper .
within large uncertainties @xcite , the picture that emerges from these studies is of star formation that has spanned a hubble time , likely to be declining over time following an early epoch of high star formation rate ( sfr ) .
the sfr has certainly declined with time over the past @xmath212 gyr , despite the large reservoir of neutral gas . in its extended star formation history
, pegdig appears to have more in common with the dirr galaxies than with a typical dsph , consistent with its retention of neutral gas to the present day and with the assertion of @xcite that transition type galaxies represent the low - mass / low - sfr end of the dwarf irregular population ( see also * ? ? ?
the precise star formation history over the full lifetime of the galaxy will be determined in a future paper in this series ( cole et al .
, in preparation ) .
we observed pegdig using the advanced camera for surveys wide field camera ( acs / wfc ) as part of the cycle 22 program go-13768 .
the observations , which comprise 34.3 and 37.4 kiloseconds in the f814w ( broad i band ) and f475w ( sloan @xmath25 band ) filters respectively , were made between 2326 july , 2015 .
29 orbits were allocated to the pegasus observations , split into 15 visits of one to two orbits each to facilitate the detection of short - period variable stars .
each orbit was broken into one exposure in each filter .
parallel observations at a distance of @xmath26@xmath26 were obtained simultaneously through the equivalent filters on the wide field camera 3 .
the charge - transfer efficiency corrected images were processed through the standard hst pipeline and photometry was done using the most recent version of dolphot , with its hst / acs - specific modules @xcite . extended objects and residual hot pixels were rejected based on their brightness profiles , and aperture corrections were derived based on relatively isolated stars picked from around the image .
stars that were found to suffer from excessive crowding noise ( crowding parameter @xmath271.0 ) due to partially resolved bright neighbors were rejected , leaving a sample of 247,390 well - measured stars with s / n @xmath165 .
we performed an artificial star test analysis with 50,000 stars , distributed following the cluster light within 10@xmath28 of the cluster center , in order to characterize the measurement errors and incompleteness .
the 50% completeness limits are ( m@xmath29 , m@xmath5 ) = ( 28.3 , 27.4 ) ; this does not reach the cluster msto due to crowding , although the oldest msto is reached for the field population . at these magnitude levels
the typical photometric error is @xmath100.08 mag , allowing us to resolve the old stellar sequences and make comparisons to isochrone models with a high degree of precision .
the co - added image built from our f814w images is shown in figure [ fig - image]a .
this tends to highlight the foreground stars and many background galaxies , along with the smoothly varying light of the intermediate - age and old field population of pegdig .
there is little to no evidence for associations of luminous , young stars , consistent with the evidence for a very low rate of massive star formation @xcite .
the distribution of bright stars is clumpy and irregular , with a few modest dust lanes that could contribute to differential reddening .
ddo216-a1 is highlighted within a 45-arcsecond box in figure [ fig - image]a .
[ fig - image]b expands this highlighted region in false color .
the gap between wfc1 and wfc2 chips runs across the bottom third of the cluster image .
the cluster is prominent in the image , even seen in projection against the densest part of the galaxy field population .
the field population is patchy and irregular , with evidence for differential reddening and an increase in the density of bright , blue stars towards the upper right .
the cluster appears to be around 810 arcseconds ( 3544 pc ) in diameter , and is nearly circular in projection .
ddo216-a1 is neither notably blue nor red compared to its surroundings .
the neutral color indicates immediately that the cluster light is dominated by first - ascent red giant branch stars , and not young main - sequence stars or asymptotic giant branch stars .
the cluster structural and photometric properties are presented in section [ sec : clusphot ] .
we present a cmd of the central 1 arcminute of the galaxy ( including the cluster ) in figure [ fig - cmd ] to highlight the differences between the galaxy field and the cluster . compared to its surroundings
, the cluster suffers from a much higher degree of crowding , but is still complete to more than two magnitudes below the level of the horizontal branch .
a 30 myr parsec isochrone @xcite chosen to approximately match the mean metallicity of pegdig ( z = 0.001 ) has been overlaid on the field cmd to indicate the distance , 900 @xmath130 kpc , and reddening , e(b@xmath0v ) = 0.16 @xmath10.02 mag to the galaxy ( see below ) .
the field stars contain a strong population of red clump stars , a relatively broad red giant branch , and main sequence that is well - populated across a wide range of magnitudes , indicating a wide range of stellar ages .
the cluster shows no enhancement in density of red clump stars , asymptotic giant branch ( agb ) stars , or upper main sequence stars compared to the field , but shows a well - populated horizontal branch ( hb ) and red giant branch ( rgb ) , with a mostly red horizontal branch morphology .
the main - sequence turnoff ( msto ) of the cluster is not obvious due to the increased crowding ; if the cluster were younger than @xmath278 gyr we would see its msto above the crowding limit of the data .
the tip of the red giant branch in the cluster terminates at m814w @xmath221 , very similar to that of the field .
the first ccd images of pegdig were obtained by @xcite , who noted three star cluster candidates and also commented on the high number of background galaxies visible around and through the galaxy .
indeed , pegdig is not far from the supergalactic plane ; it appears in projection on the sky in the foreground of a galaxy group in the outskirts of the pegasus cluster ( see e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
@xcite listed their star cluster candidates as a13 ; the potential for naming confusion with the pegasus cluster of galaxies leads us to adopt hoessel & mould s nomenclature for the central star cluster in ddo216 , which we hereafter refer to as ddo216-a1 .
@xcite report the magnitude and color for cluster a1 , in the thuan - gunn system , as @xmath30 = 18.75 and @xmath31 = 0.45 , and estimate its age to be @xmath17 2 gyr from the integrated color .
the first hubble space telescope observations of pegdig were reported by @xcite , who placed the wfpc2 camera in the central parts of the galaxy such that the pc chip barely included the outskirts of ddo216-a1 .
the cluster is also clearly identifiable in their 06-seeing wiyn 3.5 m telelecope images .
these authors note that the cluster is only moderately dense , with a rather large diameter of @xmath240 pc , and that although it is located very centrally within pegdig it falls well short of being a galaxy nucleus or super star cluster @xcite .
the shallowness of their cmds and the positioning of the cluster mainly outside the wfpc2 field prevented any further quantitative work although they confirm the claim of @xcite that the cluster is most likely older than 2 gyr , based on its lack of bright main - sequence stars @xcite .
subsequently the cluster appears to have been lost to the literature , failing to be included in compilations of the properties of local group galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ) , papers specifically devoted to the statistics of star clusters in dwarf galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and further space- and ground - based studies of pegdig ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
given the recent surge in interest in the nature of extended star clusters in dwarf galaxies and in the physical differences between clusters and galaxies , the presence of an understudied , luminous , high surface brightness star cluster at the very center of one of the local group s faintest photographically discovered galaxies is remarkable .
ddo216-a1 is located at right ascension 23:28:26.3 , declination @xmath3214:44:25.2 ( j2000.0 ) .
this is just 6@xmath28 ( 26 pc ) south of the center of the outer red light isophotes of pegdig as measured by @xcite .
the cluster is of high surface brightness and has a very distinctly identifiable core , but determination of the cluster properties is complicated by the fact that it is embedded in the highest surface brightness part of pegdig , and by the fact that significant differential reddening appears to be present .
the differential reddening can be seen both in the color spread of red giants and upper main sequence stars , and in the presence of dust lanes in the false color acs / wfc images .
ddo216-a1 is nearly circular ; using the iraf _ stsdas _ task _ ellipse _ , we find an ellipticity @xmath33 @xmath34 ( 1@xmath35 ) @xmath360.15 .
we calculated the half - light radius of the cluster using star counts of horizontal branch stars .
the pegdig field has a compound hb morphology , with both red and blue hb stars present , spread nearly evenly across the acs / wfc field .
the cluster core , within 40 pixels of the center , has roughly an order of magnitude higher surface density of hb stars , so these make a convenient measure for estimating the cluster extent with the minimum field contamination . in a region from 1215@xmath28 outside the cluster , the density of hb stars is 0.14 arcsec@xmath3 .
the hb density rises to twice this value at a radius of 8@xmath45 from the cluster center , which we take to be the area over which the cluster is clearly dominant over the field . within this 8@xmath45 ( @xmath237 pc ) circle
, we would expect to find 32 @xmath16 hb stars from the field population alone ; the actual cmd of this area yields 89 hb stars .
subtracting the field contribution and finding the radius within which half the cluster is contained , a half - light radius of 3@xmath41 @xmath1 0@xmath37 is derived ( 13.4 @xmath1 1.3 pc ) .
this result does not significantly change if the upper red giant branch stars are included , and the contamination by field stars is higher .
we measured the flux in concentric apertures around the cluster , using annuli with outer radii from 0@xmath4820@xmath28 in size , and an annulus from 2530@xmath28 to measure the field contribution .
we measured the f814w and f475w surface brightness profiles and found them to be consistent with the half - light radius derived from hb star counts .
we merged the two bandpasses into a single profile by simultaneously fitting a king model with a common core radius r@xmath8 and concentration index @xmath9 @xmath34 log(r@xmath38/r@xmath8 ) and two central surface brightness values , one for the f814w image and one for the f475w image .
the resulting fit is shown for the f814w surface brightness in figure [ fig - profile ] .
the king model core radius is r@xmath8 = 2@xmath412 @xmath1 0@xmath491 , the concentration index is @xmath9 = 1.24 @xmath10.39 , and the central surface brightnesses are @xmath39 = 20.85 @xmath10.23 and @xmath40 = 22.68 @xmath10.24 mag arcsec@xmath3 , respectively .
the outer regions of the cluster are difficult to reliably measure because of the high field contamination .
therefore , the total magnitude of ddo216-a1 is estimated by measuring the background - subtracted flux within the half - light radius and doubling it .
the extrapolated integrated magnitudes are ( m475w , m814w ) = ( 18.64 @xmath10.07 , 17.17 @xmath10.14 ) . transformation to a standard johnson - cousins system gives ( b , i ) = ( 18.95 , 17.39 ) , and a transformed v magnitude of 18.13 .
the measured total magnitude is thus in reasonable accord with the original discovery values in @xcite .
the absolute magnitude of ddo216-a1 is m@xmath5 = @xmath07.90 @xmath10.16 for the distance and reddening derived from our cmds of the cluster and its surrounding fields .
this yields a v magnitude m@xmath41 = @xmath07.14 , slightly fainter than the peak of the milky way globular cluster luminosity function .
the color b@xmath0v = 0.82 is quite red for a globular cluster , but when dereddened [ ( b@xmath0v)@xmath42 = 0.66 ] is entirely consistent with typical metal - poor globulars . candidate variable stars were identified in the photometric catalog using the method of @xcite , as implemented and applied to _ hst _
photometry in @xcite and numerous subsequent papers ( e.g. , * ? ? ?
* and references therein ) .
the algorithm labels stars as likely variables if the photometric scatter of the individual data points is much larger than the photometric errors , the scatter is not due to a small number of outliers , and the light curve is periodic , with a best - fit period found through an application of the @xcite algorithm .
the observations are highly sensitive to variables of period from @xmath20.13 days , with 29 individual data points in each filter , spread over 2.8 days with an average gap between observations of 0.05 days . over the entire acs / wfc field ,
several hundred likely variables were flagged , strongly clustered on the horizontal branch in the cmd , with periods of @xmath20.50.6 day .
a scattering of brighter , longer period variables , mostly within the cepheid instability strip , was also identified .
the horizontal branch variables are found to be distributed evenly across the galaxy , with the exception of a strong concentration at the location of ddo216-a1 .
the field variable star population of pegdig will be analysed in a future paper ( skillman et al . , in preparation ) . here
, we identify genuine variables within 8 arcseconds of the cluster center and classify them using their light curve shapes . within this area , there are 32 variables , all on or near the horizontal branch ; the corresponding number in a typical nearby comparison region is 4 , so the vast majority of the 32 are likely to be cluster members . among the cluster variables , 27 are fundamental mode ( rrab )
pulsators , and 5 are overtone ( rrc ) pulsators .
sample light curves in each filter for one star of each variable type are shown in figure [ fig : rrlyr ] , with the period and mean magnitude shown .
four additional variables with brighter magnitude but unexceptional periods and colors were also found , @xmath20.10.4 mag above the hb .
their colors are too blue to be population ii cepheids ; if they were the descendents of a putative blue straggler population , they would also likely fall to the red of their observed position in the cmd .
these are likely to be photometric blends ; for now , we exclude them from further analysis of the cluster properties .
the 50% completeness limit is over 2 magnitudes below the hb in the cluster region , so the census of cluster variables is likely to be very close to complete ( apart from the 4 rejected blends ) .
the list of variables within 8@xmath28 of the cluster center is given in table [ tab : rrlyr ] .
lcccccr 39184 & 1543.92 & 2330.24 & [email protected] & [email protected] & [email protected] & rrab + 39321 & 1551.63 & 2397.23 & [email protected] & [email protected] & [email protected] & rrab + 38807 & 1558.91 & 2345.44 & [email protected] & [email protected] & [email protected] & rrab + 40423 & 1571.24 & 2475.00 & [email protected] & [email protected] & [email protected] & rrab
the variable stars of ddo216-a1 are overplotted on the cluster cmd in figure [ fig - cmdplus]a .
the magnitudes have been corrected for a reddening e(b@xmath0v ) = 0.16 as derived from the color of the upper main sequence , and for distance modulus of 24.77 , derived from the mean magnitude of the fundamental mode pulsators and isochrones fits to the cluster ( see below ) .
the mean period of the 27 rrab pulsators that are likely members of ddo216-a1 is 0.556 @xmath10.039 day , and the mean period of the 5 probable rrc members is 0.356 @xmath10.012 day .
this , along with the roughly 5:1 ratio of fundamental to overtone pulsators , identifies the cluster as belonging to oosterhoff type i ( oo i , * ? ? ?
most galactic globular clusters more metal - rich than [ fe / h ] @xmath2 @xmath01.7 fall into this category , with @xmath43p@xmath44 @xmath20.55 d ; most of the more metal - poor clusters are type oo ii , with @xmath43p@xmath44 @xmath20.65 d @xcite .
interestingly , many of the globular clusters in the fornax and sagittarius dsph and in the large magellanic cloud are of intermediate or indeterminate oosterhoff type ( e.g. , * ? ? ?
* ; * ? ? ?
* ) ; whatever effect causes these differences in variable populations seems inapplicable to ddo216-a1 .
ddo216-a1 appears to be quite ordinary in its variable star population ; normalised to a cluster with m@xmath18 = @xmath07.5 , the specific frequency of rr lyraes is @xmath242 .
the mean f475w and f814w magnitudes of the probable rrab cluster members are 26.074 @xmath10.126 and 25.088 @xmath10.111 , respectively .
we convert the magnitudes to a mean v magnitude following the procedure in @xcite , finding @xmath43v@xmath45 = 25.699 ; the rms scatter is @xmath10.114 and the standard error of the mean is @xmath10.022 .
the mean rrab magnitude can be used to derive the distance , provided some estimate of the metallicity is known @xcite .
one choice is to adopt the mean metallicity and metallicity spread from the spectroscopic study of pegdig red giants by @xcite , [ fe / h ] = @xmath01.4 @xmath10.3 .
isochrone fits to the cluster red giant branch using the parsec isochrones @xcite give a slightly lower estimate , @xmath01.6 , with a random error @xmath10.2 ( see section [ sec : agemet ] ) .
the metallicity estimates are mutually consistent within the errors , so we adopt the isochrone based value since it is directly measured from the cluster and not from the field @xmath17100 parsecs away .
we use the metallicity - magnitude relationship with a zeropoint based on parallaxes of nearby field rr lyraes from @xcite to derive the variable star - based distance to ddo216-a1 . with [ fe / h]@xmath46 = @xmath01.6 @xmath10.2 and m@xmath41(rr )
= 0.214[fe / h]@xmath320.77 , we find that the distance modulus to ddo216-a1 , given the adopted reddening , is ( m@xmath0m)@xmath42 = [email protected] .
the uncertainty is dominated by the uncertainty in the v - band extinction .
the corresponding distance , 899 @xmath131 kpc , is identical to within the errors with the distance to the host galaxy @xcite , although it is formally slightly smaller
. we plot the cmd of the cluster core ( r @xmath47 8@xmath28 ) in figure [ fig - cmdplus]b . by fitting the distribution of stars using synthetic cmds and artificial star tests we are able to constrain the age and metallicity of ddo216-a1 .
the cmd - fitting programs , _ forge / anneal _ , are described in @xcite , and numerous other references .
we adopt the most recent set of isochrones from the parsec family of models @xcite .
these models have been calculated assuming the solar metallicity [ m / h ] = 0 corresponds to a mass fraction of elements heavier than helium z = 0.01524 . in the fits , we subtracted off a scaled hess diagram drawn from an annulus around the cluster to account for field contamination .
the best - fit distance from the cmd is 24.80 @xmath10.10 , entirely consistent with the value derived from the variable stars .
the reddening value is equal to the mean reddening of the upper main sequence stars in the nearby field , e(b@xmath0v ) = 0.16 @xmath10.02 .
we therefore choose to reduce the variance in our fits by holding these values fixed in our final determination of the cluster age and metallicity .
because of the high degree of crowding , the cluster cmd becomes incomplete significantly brighter than the msto .
there is thus some age / metallicity / reddening degeneracy in the best - fit solutions , as there always must be when the fit information is dominated by red giant branch stars .
the best - fit isochrones have z = 0.0004 @xmath10.0002 ( [ m / h ] = @xmath01.6 @xmath10.2 ) , with age = 12.3 @xmath10.8 gyr ( random error only ) .
covariance between age and metallicity mean that good models on the younger side of the range tend to have higher metallicity , and vice versa . given the field contamination and crowding we are unable to rule out small numbers of younger stars or a cluster metallicity spread without spectroscopic information .
figure [ fig - cmdplus]b shows three isochrones overplotted from the range of acceptable solutions .
the best - fit age and metallicity is plotted in blue , while the purple track shows z = 0.0024 ( [ m / h ] = @xmath01.8 ) , age = 11.5 gyr , and the orange track gives z = 0.0006 ( [ m / h ] = @xmath01.4 ) , age = 13.2 gyr .
all three solutions clearly are reasonably good matches to the data , but the central part of the range gives the best reproduction of the cluster hb color .
detailed exploration of the field star cmd is beyond the scope this paper .
however , we briefly consider the ancient field star populations as a point of comparison to the cluster . in figure
[ fig - cmdplus]c we show the cmd of a region of the field far from the high surface brightness `` bar '' of pegdig , in the southern corner of the acs / wfc chip .
these stars are all more than 1@xmath488 ( 500 pc ) from ddo216-a1 , a very low surface brightness part of the galaxy completely lacking in stars @xmath102 gyr old .
this field , which has reddening consistent with e(b@xmath0v ) = 0.07 @xcite , clearly shows a metal - poor , very old population similar to the globular cluster .
the predominantly red hb morphology suggests a slightly more metal - rich population , consistent with the spectroscopic metallicity measurements from @xcite .
main sequence stars with m814w @xmath2 @xmath3212 , and the hint of a `` vertical red clump '' ( vrc ; * ? ? ?
* ) can be seen in fig .
[ fig - cmdplus]c , indicating either a small population of intermediate - age , metal - poor stars , or ancient blue stragglers and their descendents .
the field blue stragglers and vrc give a clue about where to look in the cluster cmd for similar types of star ; unfortunately both regions in the cmd of ddo216-a1 are completely dominated by field populations of intermediate age .
the field vrc is at a color of @xmath21.1 and a magnitude of @xmath2 @xmath00.75 , which is inconsistent with the location of the anomalously bright variable stars in ddo216-a1 ( fig .
[ fig - cmdplus ] ) .
thus the bright cluster variables are not consistent with contamination by metal - poor intermediate age field stars or blue stragglers .
although we can not come to a definite conclusion about their nature , it seems likely that they are indeed photometric blends .
lc right ascension ( j2000.0 ) & 23:28:26.3 + declination ( j2000.0 ) & @xmath3214:44:25.2 + projected dist . from galaxy center & 6@xmath28 , 26 pc + m814w ( mag ) & 18.64 @xmath10.07 + m814w@xmath0m475w ( color ) & 1.47 @xmath10.16 + distance modulus , m@xmath0m@xmath42 & 24.77 @xmath10.08 + reddening , e(b - v ) & 0.16 @xmath10.02 + m814w ( mag ) & @xmath07.90 @xmath10.16 + ( m475w@xmath0m814w)@xmath42 & 1.18 @xmath10.16 + central surface brightness ( m@xmath5 arcsec@xmath3 ) & 20.85 @xmath10.17 + half - light radius , r@xmath49 & 3@xmath41 , 13.4 pc + core radius , r@xmath50 ( arcsec ) & 2.1 @xmath10.9 + concentration index ( log(r@xmath51/r@xmath50 ) & 1.24 @xmath10.39 + metallicity ( [ m / h ] ) & @xmath01.6 @xmath10.2 + age ( gyr ) & 12.3 @xmath10.8 + log(m / m@xmath13 ) & 5.0 @xmath10.1 + # rr lyrae stars & @xmath230 + mean rrab period ( d ) & 0.556 @xmath10.039 + mean rrab magnitude ( v ) & 25.70 @xmath10.11
the photometric and structural parameters of ddo216-a1 are summarised in table [ tab : props ] . by every available measure
, ddo216-a1 is a bona fide , ancient globular cluster , indistinguishable in many ways from the globular clusters of the milky way and large magellanic cloud .
while the cluster was first identified over 35 years ago , it has only sporadically been recorded in the literature @xcite . with an absolute magnitude m@xmath41 @xmath2 @xmath07.1
, ddo216-a1 contributes roughly 0.5% of the v - band light of pegdig this makes its lack of study all the more remarkable , given the long history of observational studies of pegdig .
the reasons for the omission of ddo216-a1 from lists of globular clusters in dwarf galaxies probably stem from a combination of the location of ddo216-a1 , seen in projection against the densest part of pegdig ; the unusually extended nature of the cluster , which both gives it much less contrast with its surroundings and makes it somewhat unlike galactic globular clusters ; and the prevalence of background galaxies in the field . at a distance of 900 kpc
, it requires diffraction limited imaging to resolve the cluster even to the level of the horizontal branch .
the cluster is not prominently visible in spitzer space telesope irac images @xcite due to its lack of asymptotic giant branch stars ; with hindsight , it is obvious in sdss images , although unresolved . in the absence of spectroscopic information , the cluster mass can only be estimated in a model - dependent way or by comparison to better - studied , similarly luminous clusters . @xcite compared the dynamical mass - to - light ratios for a large set of milky way and magellanic cloud globular clusters to predictions of population synthesis models by @xcite .
using their preferred initial mass function @xcite , they found a mean model m / l @xmath21.9 for the old clusters .
the median dynamical mass to light ratio for the clusters in their sample was 82@xmath17% of this value , with a substantial scatter .
our absolute magnitude for ddo216-a1 is m@xmath18 = @xmath07.14 @xmath10.16 , which translates to a v - band luminosity of ( 5.97 @xmath10.95)@xmath52 l@xmath13 .
using the m / l estimates from @xcite gives either [email protected] ( population synthesis ) or [email protected] ( dynamical ) @xmath5310@xmath7 m@xmath13 .
the true range of possible values is even larger , because of potential variations in the initial mass function and the observed variations between clusters . given the lack of kinematic constraints , an appropriate way to express the probable mass of the cluster is @xmath54(m / m@xmath13 ) = [email protected] .
this is entirely consistent with mass estimates for similarly bright and extended , old clusters in the milky way ( e.g. , ic4499 , * ? ? ?
* ) , and the lmc ( e.g. , reticulum , * ? ? ?
the cluster s projected position near the center of pegdig raises the question of its provenance and survival . if the cluster formed in situ at the center of the galaxy , then it is natural to ask how it has survived tidal evaporation for 12 gyr .
alternatively , if ddo216-a1 formed at an arbitrary location in the galaxy , then dynamical friction must be acting efficiently enough to bring it nearly to the center within its lifetime .
survivability of clusters in dwarf galaxies can be calculated probabilistically using analytical models for dynamical friction and cluster evolution ( leaman et al . , in preparation ) . using the dynamical friction formula from @xcite , a range of galaxy mass profiles and cluster orbits
can be tested to see if there are any plausible initial conditions conducive to cluster inspiral and survival .
the half - light radius of pegdig is @xmath2700 pc @xcite and its stellar mass is log(m@xmath55 ) @xmath210@xmath56 m@xmath13 @xcite , but its mass profile is not extremely well known . to a first approximation it could be taken as similar to a scaled - down version of wlm , which is well - fit by a @xcite profile with virial mass m@xmath57 = 10@xmath58 m@xmath6 and concentration parameter @xmath9 = 15 @xcite . to reflect the uncertainties in the parameters
, we ran 2000 trials in which the important unconstrained parameters were drawn at random , and the dynamical friction and tidal destruction timescales were calculated analytically . the parameters and their range of sampled values were the initial distance and orbital eccentricity for ddo216-a1 ( evenly distributed from 02 kpc and from 01 , respectively ) , and the virial mass ( log - normal distributed around 10@xmath58 m@xmath13 ) , concentration index ( normally distributed around @xmath9 = 12.5 ) , and einasto profile slope ( evenly distributed between 01 ) for the pegdig halo . in this set of trials , 27% of the clusters
are found to have survival times longer than 12 gyr and dynamical friction timescales shorter than this . within the range of parameters considered , there were no strong trends of survivability in the concentration index , einasto profile slope , or virial mass , but the best cluster survivability is found for birthplaces from @xmath2300 - 1000 pc from the galaxy center ; nearly half of 10@xmath7 m@xmath13 clusters born within this range sink to within @xmath10100 pc of the center without tidal destruction over their lifetime .
clusters born interior to this region tend to be tidally disrupted , and clusters born in the galaxy outskirts have dynamical friction timescales longer than a hubble time .
the general feature of the analysis , that in many cases clusters will be disrupted , but that the most massive will sometimes survive to be observed near the center of the host galaxy , is consistent with advanced numerical simulations of star formation in dwarfs ( governato , private communication ) .
these analytical calculations show that the cluster location is consistent with formation across a large volume of pegdig , excluding the central region where it is now observed .
given the propensity of clusters to dissolve when located at the center of the galaxy , it seems unlikely that ddo216-a1 formed at its current location .
because dynamical friction tends to stall when the cluster reaches the radius at which the host galaxy density profile flattens into a core , it is not surprising that the cluster is not observed at the precise center of pegdig .
both of these factors point to the likelihood that the true distance from the pegdig center to the cluster is larger than the projected separation .
a census of globular clusters in local group dwarf galaxies has been given in @xcite ; the count has updated slightly since then as the result of hst and wide - field surveys of dwarfs at large distances , at low surface brightness , or where foreground confusion is high . however , most of the additions are small clusters at least an order of magnitude less luminous than ddo216-a1
. notable exceptions to this trend are the brightest 4 of the 7 new clusters discovered in ngc 6822 by @xcite and @xcite ; ngc 6822-sc1 is of similar age , half - light radius , and luminosity to ddo216-a1 , although it is significantly more metal - poor @xcite .
three local group galaxies fainter than pegdig host star clusters : eri ii ( m@xmath41 = @xmath07.1 , * ? ? ?
* ) , and xxv ( m@xmath41 = @xmath09.7 , * ? ? ?
* ) , and and i ( m@xmath41 = @xmath011.7 , * ? ? ?
each hosts a single cluster , but all are far fainter than ddo216-a1 ( m@xmath41 @xmath2 @xmath03.5 to @xmath05 ) , and are quite diffuse and extended by comparison .
these galaxies are all dsphs ; the cluster ddo216-a1 is as luminous as the entire galaxy eri ii .
the cluster - hosting dirr galaxies ngc 6822 ( m@xmath41 = @xmath015.2 ) and wlm ( m@xmath41 = @xmath014.2 ) are both far more luminous than pegdig .
the lowest - luminosity local group galaxies with recorded globular clusters as bright as ddo216-a1 are the fornax and sagittarius dwarf spheroidals @xcite .
both galaxies are @xmath21 mag brighter than pegdig and each has multiple clusters .
fornax hosts 5 globulars , and sagittarius at least 4 @xcite , possibly as many as 9 @xcite .
normalising each galaxy to an absolute magnitude of m@xmath41 = @xmath015 , the specific frequency of globular clusters is 29 for fornax , 59 for sagittarius , 7 for ngc 6822 , 2 for wlm , and 10 for pegdig .
pegdig thus has a rather ordinary specific cluster frequency compared to other local group galaxies and its role as cluster host is not surprising .
statistics of clusters in dwarf galaxies beyond the local group are necessarily less complete , but summaries of statistical properties of clusters in dwarfs can be found in , e.g. , @xcite and references therein .
ddo216-a1 appears to be a typical old and metal - poor cluster of the type common to both giant and dwarf galaxies , although it is unusual to find a cluster as luminous as ddo216-a1 in a galaxy as faint as pegdig .
the contribution of the cluster to the total light of the galaxy is @xmath20.5% ; @xcite has shown that globular clusters in dwarf galaxies can contribute up to 2025% of the metal - poor stars in a given dwarf . while a detailed analysis awaits the full sfh of pegdig , it appears as though the fraction is much smaller in this case . unlike fornax , wlm , and
ngc 6822 , the cluster does not have a dramatically lower metallicity than the galaxy as a whole , only about 0.3 dex less .
the structural parameters of the cluster resemble those of the `` faint fuzzy '' clusters found in lenticular galaxies @xcite , but its color is much bluer and it otherwise appears typical of the globular cluster population of dwarfs .
ddo216-a1 is unusually close to the center of its host galaxy , and is also unusually extended for a cluster with small projected galacticentric distance @xcite .
however , it clearly falls far below the luminosity and surface brightnesss of clusters that are typically identified as galactic nuclei @xcite .
@xcite finds that among dsphs with globular clusters , more than half show clusters seen in projection against the center of the galaxy .
while ddo216-a1 fits among these clusters in luminosity , its large radius distinguishes it from the ( usually ) compact central clusters .
however , it is still quite a bit more compact than very extended clusters like scl - de1 gc1 , with a half - light radius of 22 pc @xcite .
pegdig is the lowest - mass gas - rich galaxy in the local group with a major star cluster .
pegdig is a transition type galaxy , with characteristics of both irregular and spheroidal galaxies ; its principal dsph - like quality is the very low rate of current star formation @xcite . in the ( r@xmath59 , m@xmath41 ) plane , ddo216-a1 is more typical of clusters in dsph galaxies than in dirr @xcite .
if pegdig is considered to be a dirr , then ddo216-a1 would be one of the largest clusters known in a small dirr .
more extreme examples are very rare .
for example , in the list of @xcite , there is only one less - luminous dirr with a comparably bright cluster , the m@xmath41 = @xmath011.9 field galaxy d634 - 03 , at a distance of 9.5 mpc . by comparison ,
there are numerous cases of globulars in dsphs at the same host galaxy absolute magnitude .
the overall trend with galaxy morphology appears to suggest that globular cluster populations are typically much poorer among the gas - rich dwarfs , which also lack nuclear clusters .
ddo216-a1 is more extended than @xmath290% of galactic and magellanic cloud globular clusters , although given its absolute magnitude and half - light radius it is not an extreme outlier .
it would not be out of place among `` outer halo '' globulars such as ic 4499 or ngc 5053 , magellanic cloud clusters like reticulum ( lmc ) , or clusters like ngc 6822-sc1 .
contrastingly , it is much more luminous than the extended clusters fornax 1 , arp 2 , ter 8 , or pal 12 , similarly extended clusters that are either definite or probable members of spheroidal galaxies . given the complexity of the field star background , it is impossible to say whether there are multiple ages or metallicities present in the cluster without spectroscopy .
these features would be indicative of cluster mergers or in situ star formation from newly accreted gas in the cluster , either of which could contribute to the formation of a nuclear star cluster .
however , circumstantial evidence argues against it being a nuclear cluster : it would be an extreme outlier from the host mass - cluster radius and -cluster mass relations presented in @xcite , being both too extended for its mass and too massive for its host compared to other nuclear star clusters .
we have imaged the central , extended star cluster in the pegasus transition dwarf with hst / acs , and obtained a color - magnitude diagram reaching @xmath20.5 mag above the cluster main - sequence turnoff .
ddo216-a1 is in some respects a typical globular cluster but is more extended than most .
we find in particular the following major features : 1 .
we have confirmed that ddo216-a1 is a bona fide globular cluster , with absolute i - magnitude m@xmath5 = @xmath60 , a mass of @xmath1110@xmath7 m@xmath13 , and a half - light radius @xmath61 pc . while it is larger than @xmath290% of milky way globular clusters , it is not structurally dissimilar to some of the outer halo milky way globular clusters , and is not an extreme example of the type of extended cluster that seems to be characteristic of some dwarf galaxies . 2 . based on the color - magnitude diagram ,
the cluster is ancient , 12.3 @xmath10.8 gyr old , and is moderately metal - poor , [ fe / h ] = @xmath01.6 @xmath10.2 . in its variable star properties , ddo216-a1 harbors @xmath230 rr lyrae stars , and
is an oosterhoff type i cluster with a specific frequency s @xmath242 .
3 . ddo216-a1 contributes @xmath20.5% of the v - band luminosity of pegdig , but the galaxy does not have an anamolously high specific frequency of clusters compared to other dwarf galaxies . despite the low mass of the host ,
the cluster is very close to the peak of the globular cluster luminosity function for all local group galaxies .
the cluster is seen in projection within 30 pc of the galaxy center , but it is much fainter than dwarf galaxy nuclei .
birth at a distance of @xmath110.31 kpc and subsequent infall due to dynamical friction is a possible scenario resulting in the observed cluster position . because dynamical friction is inefficient within cored mass distributions , and the cluster has not been tidally disrupted , it is likely that the true distance between the cluster and galaxy center is a few times larger than the projected separation .
the association of globular clusters with intense episodes of star formation involving very large gas masses @xcite indicates that pegdig might have had a tumultuous early history . the observed relationship between the size of the largest star cluster and the peak sfr ( e.g. , * ? ? ?
* ; * ? ? ?
* ) suggests that pegdig should have experienced its highest sfr around the time that ddo216-a1 formed .
considering the large star formation surface densities associated with the formation of a cluster as massive as ddo216-a1 ( e.g. , * ? ? ?
* ) and the constraints on the total stellar mass of the galaxy , this raises the likelihood that pegdig may have formed a large fraction of its stars in an intense burst around the time of cluster formation .
the resulting peak in sfr at early times would tend to make pegdig more similar to a prototypical dsph than to a dirr .
although pegdig is not currently close to any other known system , its radial velocity and distance from m31 suggest that it is not likely to be on its first infall into the local group .
@xcite have found in their cosmological simulations of structure formation that in mock local groups , the majority of dwarf galaxy - sized subhaloes found within 11.5 virial radii of a large halo at redshift @xmath15 = 0 have previously spent time inside the virial radius . timing arguments using numerical action reconstructions ( e.g. , * ? ? ?
* ) , while subject to significant uncertainty due to unknown initial conditions , also suggest that pegdig has not always been isolated .
close encounters with m31 or another dwarf could have dramatically increased the sfr and therefore the statistical probability for a large cluster to be formed .
@xcite predict from cosmological simulations that most dwarf - dwarf major mergers tend to occur near the time of first infall into the virial radius of a larger parent galaxy , consistent with the large age of ddo216-a1 .
the next paper in this series will examine the complete star formation history of pegdig ; the sample cmd for an outer field indicates that photometry well below the oldest msto will allow a detailed reconstruction of the sfh back to the oldest ages .
support for this work was provided by nasa through grant number hst go-13768 from the space telescope science institute , which is operated by aura , inc .
, under nasa contract nas5 - 26555 .
mbk acknowledges support from nsf grant ast-1517226 and from nasa grants hst - ar-12836 , hst - ar-13888 , and hst - ar-13896 awarded by stsci .
aac is grateful to chris howk , jay gallagher , and warren hankey for help chasing obscure references .
this research made extensive use of nasa s astrophysics data system bibliographic services .
aparicio , a. , gallart , c. , & bertelli , g. 1997 , , 114 , 669 benedict , g.f . , et al .
2011 , , 142 , 187 bernard , e.j . , monelli , m. , gallart , c. , et al .
2009 , , 699 , 1742 billett , o.h . , hunter , d.a . , & elmegreen , b.g .
2002 , aj , 123 , 1454 boyer , m.l . ,
skillman , e.d . ,
van loon , j.th . ,
gehrz , r.d . ,
& woodward , c.e .
2009 , , 697 , 1993 boylan - kolchin , m. , weisz , d.r . ,
johnson , b.d . ,
bullock , j.s . ,
conroy , c. , & fitts , a. 2015 , , 453 , 1503 boylan - kolchin , m. , weisz , d.r . ,
bullock , j.s . , & cooper , m.c .
2016 , , 462l , 51 bressan , a. , marigo , p. , girardi , l. , salasnich , b. , del cero , c. , rubele , s. , & nanni , a. 2012 , , 427 , 127 brodie , j.p . , & larsen , s.s .
2002 , , 124 , 1410 brodie , j.p . , & strader , j. 2006 , , 44 , 193 brodie , j.p . ,
romanowsky , a.j . , strader , j. , & forbes , d.a .
2011 , , 142 , 199 bruzual , g. & charlot , s. 2003 , , 344 , 1000 caputo , f. , castellani , v. , & deglinnocenti , s. 1995 , , 304 , 365 catelan , m. , & smith , h.a .
2015 , `` pulsating stars '' , ( wiley - vch verlag : weinheim , germany ) chabrier , g. 2003 , , 115 , 763 chincarini , g. , & rood , h.j . 1976 , , 88 , 388 cole , a.a . , skillman , e.d .
, tolstoy , e. , et al .
2007 , , 659 , l17 cole , a.a . ,
weisz , d.r . ,
dolphin , a.e .
, et al .
2014 , , 795 , 54 cook , d.o . , seth , a.c . , dale , d.a . , et al .
2012 , , 751 , 100 crnojevi , d. , sand , d.j . , zaritsky , d. , spekkens , k. , willman , b. , & hargis , j.r .
2016 , , 824 , 14 cusano , f. , garofalo , a. , clementini , g. , et al .
2016 , , accepted , arxiv:1606.06862 da costa , g.s . , & armandroff , t.e .
1995 , , 109 , 2533 da costa , g.s . ,
grebel , e.k . ,
jerjen , h. , rejkuba , m. , & sharina , m.e .
2009 , , 137 , 4361 deason a. , wetzel , a. , & garrison - kimmel , s. 2014 , , 794 , 114 demarque , p. , zinn , r. , lee , y .- w . , & yi , s. 2000 , , 119 , 1398 dolphin , a.e . 2000 , , 112 , 1383 dolphin , a.e . ,
saha , a. , skillman , e.d . , et al .
2001 , , 550 , 554 fisher , j.r . , &
tully , r.b .
1975 , , 44 , 151 forbes , d.a . , masters , k.l . , minniti , d. , & barmby , p. 2000 , , 358 ,
471 gallagher , j.s . ,
tolstoy , e. , dohm - palmer , r.c . ,
skillman , e.d . ,
cole , a.a . ,
hoessel , j.g . ,
saha , a. , & mateo , m. 1998 , , 115 , 1869 garrison - kimmel , s. , boylan - kolchin , m. , bullock , j.s . , & lee , k. 2014 , , 438 , 2578 georgiev , i.y . ,
puzia , t.h . ,
hilker , m. , & goudfrooij , p. 2009a
, , 392 , 879 georgiev , i.y . , hilker , m. , puzia , t.h . , goudfrooij , p. , & baumgardt , h. 2009b , , 396 , 1075 georgiev , i.y . , bker , t. , leigh , n. , ltzgendorf , n. , & neumayer , n. 2016 , , 457 , 2122 grebel , e.j . , dolphin , a.e . , & guhathakurta , p. 2000 , astron .
gesellschaft abstract ser . , 17 , p61 hankey , w.j . & cole , a.a .
2011 , , 411 , 1536 holmberg , e. 1958 , medd .
lunds astr .
ii , 136 hoessel , j.g .
, & mould , j.r .
1982 , , 254 , 38 huxor , a.p . ,
ferguson , a.m.n .
, veljanoski , j. , mackey , a.d . , & tanvir , n.r .
2013 , , 429 , 1039 hwang , n. , lee , m.g . ,
lee , j.c . , park , w .- k . ,
park , h.s . , kim , s.c . , park , j .- h .
2011 , , 738 , 58 jackson , d.c . ,
cannon , j.m . , skillman , e.d . ,
lee , h. , gehrz , r.d . ,
woodward , c.e .
, & polomski , e. 2006 , , 646 , 192 johnson , l.c . , seth , a.c . , dalcanton , j.j . , et al .
2016 , , 827 , 33 kirby , e.n . ,
cohen , j.g . ,
guhathakurta , p. , cheng , l. , bullock , j.s . , & gallazzi , a. 2013 , , 779 , 102 kirby , e.n . ,
bullock , j.s .
, boylan - kolchin , m. , kaplinghat , m. , & cohen , j.g .
2014 , , 439 , 1015 kniazev , a.y . ,
brosch , n. , hoffman , g.l . ,
grebel , e.k . ,
zucker , d.b .
, & pustilnik , s.a .
2009 , , 400 , 2054 krienke , k. , & hodge , p.w .
2001 , , 113 , 1115 lafler , j. & kinman , t. d. 1965 , , 11 , 216 larsen , s.s .
2002 , , 124 , 1393 larsen , s.s .
2015 , in `` formation , evolution and survival of massive star clusters '' , iau symp . 316 , eds .
c. charbonnel & a. nota , arxiv:1510.03270 law , d.r . , &
majewski , s.r .
2010 , , 718 , 1128 leaman , r.c . , venn , k.a .
, brooks , a.m. , et al .
2012 , , 750 , 33 mackey , a.d . &
gilmore , g.f .
2003 , , 345 , 747 mateo , m. , , 36 , 435 mcconnachie , a.w . ,
venn , k.a . , irwin , m.j . ,
young , l.m . , & geehan , j.j .
2007 , , 671 , l33 mcconnachie , a.w .
2012 , , 144 , 1 mclaughlin , d.e . &
van der marel , r.p .
2005 , , 161 , 304 mcquinn , k.b.w . ,
skillman , e.d . , dolphin , a.e . , et al .
2015 , , 812 , 158 miller , b.w . , &
lotz , j.m 2007 , , 670 , 1074 oconnell , r.w . ,
gallagher , j.s . , & hunter , d.a .
1994 , , 433 , 650 oosterhoff , p.th .
1939 , the observatory , 62 , 104 orban , c. , gnedin , o.y . , weisz , d.r . ,
skillman , e.d . , dolphin , a.e . , & holtzman , j.a .
2008 , , 686 , 1030 navarro , j. , frenk , c.s .
, & white , s.d.m .
1997 , , 490 , 493 petts , j.a . ,
read , j.i .
, & gualandris , a. 2016 , , doi:10.1093/mnras
/ stw2011 robertson , b.e . ,
furlanetto , s.r . ,
schneider , r. , et al .
2013 , , 768 , 71 saha , a. , & hoessel , j.g .
1990 , , 99 , 97 sandage , a. 1990 , , 350 , 631 schlafly , e.f . , & finkbeiner , d.p .
2011 , , 737 , 103 sharina , m.e . ,
puzia , t.h . , & makarov , d.i .
2005 , , 442 , 85 sharina , m.e . ,
puzia , t.h . , & krylatyh , a.s .
2007 , astbu , 62 , 209 shaya , e.j . , &
tully , r.b .
2013 , , 436 , 2096 skillman , e.d . ,
bomans , d.j . , &
kobulnicky , h.a .
1997 , , 474 , 205 skillman , e.d . ,
ct , s. , & miller , b.w .
, 2003a , aj , 125 , 593 skillman , e.d . ,
tolstoy , e. , cole , a.a .
, et al . 2003b , , 596 , 253 skillman , e.d . , hidalgo , s.l .. weisz , d.r . , et al .
2014 , , 786 , 44 sollima , a. , cacciari , c. , bellazzini , m. , & colucci , s. 2010 , , 406 , 329 suntzeff , n.b . , schommer , r.a . ,
olszewski , e.w . , walker , a.r .
1992 , , 104 , 1743 tikhonov , n.a .
2006 , astl , 32 , 149 veljanoski , j. , ferguson , a.m.n . ,
mackey , a.d .
, et al .
2015 , , 452 , 320 weisz , d.r . ,
dalcanton , j.j . , williams , b.f . , et al .
2011 , , 739 , 5 weisz , d.r . , dolphin , a.e . , skillman , e.d . , holtzman , j. , gilbert , k. , dalcanton , j.j .
, & williams , b.f .
2014a , , 789 , 147 weisz , d.r . , dolphin , a.e . , skillman , e.d . , holtzman , j. , gilbert , k. , dalcanton , j.j .
, & williams , b.f .
2014b , , 789 , 148 young , l.m . , van zee , l. , dohm - palmer , r.c . , & beierle , m.e .
2003 , , 592 , 111 zaritsky , d.j . , crnojevi , d. , & sand , d.j .
2016 , , 826 l9 | we confirm that the object ddo216-a1 is a substantial globular cluster at the center of local group galaxy ddo216 ( the pegasus dwarf irregular ) , using hubble space telescope acs imaging . by fitting isochrones
, we find the cluster metallicity to be @xmath01.6 @xmath10.2 , for reddening e(b@xmath0v ) = [email protected] ; the best - fit age is [email protected] gyr .
there are @xmath230 rr lyrae variables in the cluster ; the magnitude of the fundamental mode pulsators gives a distance modulus of 24.77 @xmath10.08 identical to the host galaxy .
the ratio of overtone to fundamental mode variables and their mean periods make ddo216-a1 an oosterhoff type i cluster .
we find a central surface brightness [email protected] f814w mag arcsec@xmath3 , half - light radius of 3@xmath41 ( 13.4 pc ) , and absolute magnitude m@xmath5 = @xmath07.90 @xmath10.16 ( m / m@xmath6 @xmath210@xmath7 ) .
king models fit to the cluster give the core radius and concentration index , r@xmath8 = 2@xmath41@xmath10@xmath49 and @xmath9 = [email protected] .
the cluster is an `` extended '' cluster somewhat typical of some dwarf galaxies and the outer halo of the milky way .
the cluster is projected @xmath1030 pc south of the center of ddo216 , unusually central compared to most dwarf galaxy globular clusters .
analytical models of dynamical friction and tidal destruction suggest that it probably formed at a larger distance , up to @xmath111 kpc , and migrated inward .
ddo216 has an unexceptional cluster specific frequency , s@xmath12 = 10 .
ddo216 is the lowest - luminosity local group galaxy to host a 10@xmath7 m@xmath13 globular cluster , and the only transition - type ( dsph / dirr ) in the local group with a globular . |
[ [ section ] ] the cosmic microwave background ( cmb ) is very smooth . full
sky observations allow to expand the measured cmb temperature in spherical harmonics @xmath2 .
one can then determine the size of the anisotropies as a function of @xmath3 , with larger @xmath3 corresponding to smaller angles , and hence smaller length scales .
current cmb observation probe this power spectrum down to ( comoving ) length scales of about one mpc .
these observations imply very small primordial density perturbations at such large length scales , characterized by the power @xmath4 . however , it is possible that the primordial density perturbations become much larger at smaller length scales , beyond the range probed by cosmological observations .
indeed , it is conceivable that these perturbations are so large that overdense regions collapse to form primordial black holes ( pbhs ) just after the end of inflation @xcite .
they are called `` primordial '' since they do not originate from the gravitational collapse of burnt out stars ; they could thus have any mass , including masses well below or well above stellar masses . here
we assume that the fluctuations which sourced the pbhs are also generated during inflation , specifically towards the end of inflation , well after the length scales probed by conventional cosmological observations exited the horizon .
there are various constraints on pbh formation .
for example , the density of roughly stellar mass black holes has to satisfy limits from searches for microlensing .
very light black holes could have evaporated ( via hawking radiation ) in the epoch of big bang nucleosynthesis , altering the predicted isotope abundances .
these and other constraints have recently been compiled in @xcite .
they can be translated into upper limits on the amplitude of the power spectrum at the length scales relevant for pbh formation , typically @xmath5 with some scale dependence .
clearly the power spectrum has to change dramatically towards the end of inflation for pbh formation to occur . in the framework of standard slow
roll inflation , this implies that the slow roll parameters , and hence the inflaton potential , also should show large variations .
one simple , and yet well motivated , model that can feature large variations of the slow roll parameters is the running - mass model @xcite , a type of inflationary model which emerges naturally in the context of supersymmetric extensions of the standard model .
the model is of the single field type , but nevertheless it has relatively strong scale dependence ( running ) of the spectral index .
previous studies of this model @xcite showed that it can indeed accommodate sufficiently large positive running of the spectral index to allow for pbh formation .
however , the recent 7year wmap data @xcite prefer _ negative _ running of the spectral index . in this paper
we therefore revisit the running mass model .
we expand on refs.@xcite by including the running of the running of the spectral index , which is only weakly constrained by existing cmb data .
similarly , we include the running of the running of the inflaton mass parameter in the potential .
we show that for some ( small ) range of parameters , this model can still accommodate sufficiently large density perturbations near the end of inflation to allow for the formation of pbhs with mass larger than @xmath6 g , which are sufficiently long lived to be candidates for cold dark matter ( cdm ) . as a by product we show that even with the additional term in the potential , the model does not allow for significant _ negative _ running of the spectral index .
the remainder of this paper is organized as follows : in section 2 we briefly review the formalism of pbh formation . in section 3
we discuss the running mass inflation model , including the running of the running of the mass .
we compute the slow roll parameters which allow us to determine the spectral index , its running , and the running of the running . in section 4
we perform a numerical analysis where we compute the spectral index at pbh scales exactly , by numerically solving the equation of motion of the inflaton field .
finally , we conclude in section 5 .
[ [ section-1 ] ] pbhs may have formed during the very early universe , and if so can have observational implications at the present epoch , e.g. from effects of their hawking evaporation @xcite for masses @xmath7 g , or by contributing to the present `` cold '' dark matter density if they are more massive than @xmath6 g @xcite .
( these pbhs would certainly be massive enough to be dynamically `` cold '' ) .
the traditional treatment of pbh formation is based on the press - schechter formalism @xcite used widely in large
scale structure studies . here
the density field is smoothed on a scale @xmath8 . in the case at hand ,
@xmath8 is given by the mass enclosed inside radius @xmath9 when @xmath9 crossed the horizon .
the probability of pbh formation is then estimated by simply integrating the probability distribution @xmath10 over the range of perturbations @xmath11 which allow pbh formation : @xmath12 , where the upper limit arises since very large perturbations would correspond to separate closed baby universes @xcite .
we will show that in practice @xmath10 is such a rapidly decreasing function of @xmath11 above @xmath13 that the upper cutoff is not important .
the threshold density is taken as @xmath14 , where @xmath15 is the equation of state parameter describing the epoch during which pbh formation is supposed to have occurred @xcite . here
we take @xmath16 , characteristic for the radiation dominated epoch which should have started soon after the end of inflaton .
however the correct value of the threshold @xmath13 is quite uncertain .
niemeyer and jedamzik @xcite carried out numerical simulations of the collapse of the isolated regions and found the threshold for pbh formation to be @xmath17 .
we will show that pbhs abundance is sensitive to the value of @xmath13 . the fraction of the energy density of the universe in which the density fluctuation exceeds the threshold for pbh formation when smoothed on scale @xmath8 , @xmath18 , which will hence end up in pbhs with mass @xmath19 , of the horizon mass corresponding to the smoothing scale .
this is not strictly true .
in general the mass of pbhs is expected to depend on the amplitude , size and shape of the perturbations @xcite .
] is given as in press
schechter theory by to infinity . ]
@xmath20 here @xmath21 is the probability distribution function ( pdf ) of the linear density field @xmath11 smoothed on a scale @xmath9 , and @xmath22 is the fraction of the total energy within a sphere of radius @xmath9 that ends up inside the pbh . for gaussian fluctuations , the probability distribution of the smoothed density field is given by .
however , we think it is more transparent to consider @xmath23 to be the pdf of @xmath11 , which is just an integration variable in eq.([fofm ] ) . eq.([pofr ] ) shows that the functional form of @xmath24 depends on the parameter @xmath9 , which in turn depends on the horizon mass @xmath25 . ]
@xmath26 this pdf is thus uniquely determined by the variance @xmath27 of @xmath11 , which is given by @xmath28 in order to compute the variance , we therefore have to know the power spectrum of @xmath11 , latexmath:[$\mathcal{p}_{\delta}(k ) \equiv k^3/ ( 2 \pi^2 ) \left
\langle transform of the window function used to smooth @xmath11 , @xmath30 .
it is not obvious what the correct smoothing function @xmath30 is ; a top hat function has often been used in the past , but we prefer to use a gaussian window function , @xmath31 finally , on comoving hypersurfaces there is a simple relation between the density perturbation @xmath11 and the curvature perturbation @xmath32 @xcite : @xmath33 the density and curvature perturbation power spectra are therefore related by @xmath34 the mass fraction of the universe that will collapse into pbhs can now be computed by inserting eqs.([pofk ] ) and ( [ w ] ) into eq.([sigma ] ) to determine the variance as function of @xmath9 .
this has to be used in eq.([pofr ] ) , which finally has to be inserted into eq.([fofm ] ) . since we assume a gaussian @xmath24 in eq.([pofr ] ) , the integral in eq.([fofm ] ) simply gives an error function . in order to complete this calculation
one needs to relate the mass @xmath25 to the comoving smoothing scale @xmath9 . the number density of pbhs formed during the reheating phase just after the end of inflation will be greatly diluted by the reheating itself .
we therefore only consider pbhs which form during the radiation dominated era after reheating is ( more or less ) complete .
the initial pbhs mass @xmath35 is related to the particle horizon mass @xmath25 by ] @xmath36 when the scale enters the horizon , @xmath37 . here
the coefficient @xmath22 , which already appeared in eq.([fofm ] ) , depends on the details of gravitational collapse .
a simple analytical calculation suggest that @xmath38 during the radiation era @xcite . during radiation domination @xmath39 , and expansion at constant
entropy gives @xmath40 @xcite ( where @xmath41 is the number of relativistic degrees of freedom , and we have approximated the temperature and entropy degrees of freedom as equal ) .
this implies that @xmath42 where the subscript `` eq '' refers to quantities evaluated at matter radiation equality . in the early universe
, the effective relativistic degree of freedom @xmath43 is expected to be of order @xmath44 , while @xmath45 and @xmath46 @xcite ) .
the horizon mass at matter - radiation equality is given by @xmath47 where @xmath48 and ( assuming three species of massless neutrinos ) @xmath49 .
then it is straightforward to show that @xmath50 note that @xmath51 , not @xmath52 as one might naively have expected . recall that @xmath53 is related to the horizon mass at the time when the comoving scale @xmath9 again crossed into the horizon .
larger scales re enter later , when the energy density was lower ; this weakens the dependence of @xmath53 on @xmath9
. moreover , the lightest black holes to form are those corresponding to a comoving scale that re enters the horizon immediately after inflation . the gaussian window function in eq.([sigma ] )
strongly suppresses contributions with @xmath54 . at the same time
, the factor @xmath55 in eq.([pofk ] ) suppresses contributions to the integral in eq.([sigma ] ) from small @xmath56 . as a result ,
this integral is dominated by a limited range of @xmath57values near @xmath58 . over this limited range
one can to good approximation assume a power law primordial power spectrum with _ fixed _
power @xmath59 does not depend on @xmath56 ; however , @xmath59 does depend on @xmath9 , since a large range of values of @xmath9 has to be considered for pbh formation of different masses . ]
@xmath60 , with @xmath61 . with this ansatz
, the variance of the primordial density field at horizon crossing is given by @xmath62 \,,\ ] ] for @xmath63 .
the power @xmath64 is known accurately at cmb scales ; for example , @xmath65 at the `` cobe scale '' @xmath66 mpc@xmath67 @xcite . in order to relate this to the scales relevant for pbh formation , we parameterize the power spectrum as @xmath68 it is important to distinguish between @xmath69 and @xmath70 at this point .
@xmath69 describes the _ slope _ of the power spectrum at scales @xmath71 , whereas @xmath70 fixes the _ normalization _ of the spectrum at @xmath72 .
the two powers are identical if the spectral index is strictly constant , i.e. if neither @xmath59 nor @xmath73 depend on @xmath9 . however , in this case cmb data imply @xcite that @xmath74 is close to unity .
eqs.([sigmar ] ) and ( [ eqn ] ) then give a very small variance , leading to essentially no pbh formation .
significant pbh formation can therefore only occur in scenarios with running spectral index .
we parameterize the scale dependence of @xmath73 as @xcite : @xmath75 recall that we are interested in @xmath76 , i.e. @xmath77 . the parameters @xmath78 and @xmath79 denote the running of the effective spectral index @xmath59 and the running of the running , respectively : @xmath80 eq.([nofr ] ) illustrates the difference between @xmath70 and @xmath69 .
the latter has an expansion similar to eq.([nofr ] ) , but with the usual taylor expansion coefficients , @xmath81 in front of @xmath78 and @xmath82 in front of @xmath79 .
one therefore has @xmath83 setting @xmath84 for simplicity , eq.([nsofn ] ) implies @xmath85 for @xmath86 , and @xmath87 for @xmath88 .
we will compute the variance @xmath89 , and hence the pbh fraction @xmath90 , for these two relations ; they represent extreme cases if neither @xmath78 nor @xmath79 is negative .
and two different choices of the threshold @xmath91 for the solid ( dashed ) curves . on the upper [ lower ] of two curves with equal pattern and color
we have assumed @xmath92 [ @xmath87 ] . ]
the result of this calculation is shown in figure [ fig : fig1 ] .
here we have fixed @xmath94 , and show results for two choices of the threshold @xmath13 and three choices of @xmath70 .
we see that scenarios where @xmath95 ( or smaller ) are safe in the sm , because there is no model
independent limit on @xmath90 for @xmath96 g @xcite . as noted earlier ,
pbhs contributing to dark matter today must have @xmath97 g ; at this mass , they saturate the dm relic density if @xmath98 .
describes the fraction of the energy density in pbhs at the time of their formation . since they behave like matter at all times , their fractional contribution to the energy density increases during the radiation dominated epoch , and stays essentially constant during the subsequent matter dominated epoch . ]
figure [ fig : fig1 ] shows that this requires @xmath99 for @xmath91 .
the dependence on @xmath59 is much milder .
figure [ fig : fig1 ] also illustrates a serious problem that all scenarios that aim to explain the required cdm density in terms of post inflationary pbh formation face .
we just saw that this can happen only if the spectral index @xmath73 increases significantly between the scales probed by the cmb and other cosmological observations and the scale @xmath100 pc relevant for the formation of @xmath6 g pbhs .
however , @xmath73 must then _ decrease _ rapidly when going to slightly smaller length scales , since otherwise one would _ overproduce _ lighter pbhs . for example
, successful big bang nucleosynthesis requires @xcite @xmath101 , about 12 orders of magnitude below that predicted by keeping @xmath70 fixed at the value required for having @xmath6 g pbhs as cdm candidates .
we will come back to this point at the end of our paper .
another problem is that the expansion of @xmath70 in eq.([nofr ] ) will generally only be accurate if @xmath102 is not too large .
this is the case for cosmological observations , which probe scales @xmath103 mpc .
the expansion becomes questionable for the scales probed by pbh formation .
for example , fixing @xmath66 mpc@xmath67 , eq.([r ] ) gives @xmath104 for @xmath105 g. fortunately within the framework of a given inflationary scenario this second problem can be solved by computing @xmath70 and @xmath69 exactly , rather than using the expansion ( [ nofr ] ) .
observational bounds on @xmath59 and @xmath78 can e.g. be found in ref.@xcite . by requiring that these parameters remain within the @xmath106 ranges for all scales @xmath56 between @xmath107 mpc@xmath67 and @xmath108 mpc@xmath67 ,
we find @xmath109 .
here we used the error bars derived from the analysis of wmap7 data plus data on baryonic acoustic oscillations ( bao ) , ignoring possible tensor modes ( as appropriate for small field models ) , and including independent measurements of the hubble constant as a prior ( i.e. we used the `` wmap7+bao+@xmath110 '' data set ) . here
we use the pivot scale @xmath111 mpc@xmath67 where @xmath112 and @xmath113 are uncorrelated . on the other hand , eq.([nofr ] ) shows that for @xmath88 we only need @xmath114 in order to generate sufficiently large density perturbations to allow formation of @xmath6 g pbhs .
even if we set @xmath113 equal to its central value , @xmath115 , we only need @xmath116 . including the running of the running of the spectral index
thus easily allows to accommodate pbh formation in scenarios that reproduce all current cosmological observations at large scales . of course
, this kind of model independent analysis does not show whether simple , reasonably well motivated inflationary models exist that can generate a sufficiently large @xmath79 .
one scenario that quite naturally accommodates strong scale dependence of the spectral index is the running
mass inflation model @xcite .
we now apply our results to this model .
[ [ section-2 ] ] this model , proposed by stewart @xcite , exploits the observation that in field theory the parameters of the lagrangian are scale dependent .
this is true in particular for the mass of the scalar inflaton , which can thus be considered to be a `` running '' parameter .
appearing in eq.([pot1 ] ) even if @xmath117 . ]
the running of the mass parameter can be exploited to solve the `` @xmath118problem '' of inflation in supergravity @xcite .
this problem arises because the vacuum energy driving inflation also breaks supersymmetry . in `` generic '' supergravity models
the vacuum energy therefore gives a large ( gravity mediated ) contribution to the inflaton mass , yielding @xmath119 .
however , this argument applies to the scale where susy breaking is felt , which should be close to the ( reduced ) planck scale @xmath120 gev . in the running mass model , renormalization group ( rg )
running of the inflaton mass reduces the inflaton mass , and hence @xmath121 , at scales where inflation actually happens .
there are four types of model , depending on the sign of the squared inflaton mass at the planck scale , and on whether or not that sign change between @xmath122 and the scales characteristic for inflation @xcite . the simplest running
mass model is based on the inflationary potential @xmath123 where @xmath124 is a real scalar ; in supersymmetry it could be the real or imaginary part of the scalar component of a chiral superfield .
the natural size of @xmath125 in supergravity is of order @xmath126 . even for this large value of @xmath127 , which gives @xmath128
, the potential will be dominated by the constant term for @xmath129 ; as mentioned above , running is supposed to reduce @xmath130 even more at lower @xmath131 .
the potential ( [ pot1 ] ) would lead to eternal inflation .
one possibility to end inflation is to implement the idea of hybrid inflation @xcite . to that end , one introduces a real scalar `` waterfall field '' @xmath132 , and adds to the potential the terms @xmath133 here @xmath134 and @xmath135 are real couplings , and the coefficient of the @xmath136 term has been chosen such that @xmath137 has a minimum with @xmath138 if @xmath139 .
as long as @xmath140 , @xmath132 remains frozen at the origin .
once @xmath131 falls below this critical value , @xmath132 quickly approaches its final vacuum expectation value , given by @xmath141 , while @xmath124 quickly goes to zero , thereby `` shutting off '' inflation .
however , in this paper we focus on the inflationary period itself ; the evolution of perturbations at length scales that left the horizon a few e folds before the end of inflation should not be affected by the details of how inflation is brought to an end . during inflation ,
the potential is thus simply given by eq.([pot1 ] ) .
here @xmath142 is obtained by integrating an rg equation of the form @xmath143 where @xmath144 is the @xmath145function of the inflaton mass parameter . if @xmath127 is a pure susy breaking term , to one loop @xmath146 can be schematically written as @xcite @xmath147 and the second term from the yukawa interaction @xmath148 .
@xmath149 and @xmath150 are positive numbers of order one , which depend on the representations of the fields coupling to @xmath124 , @xmath151 is a gaugino mass parameter , while @xmath152 is the scalar susy breaking mass squared of the scalar particles interacting with the inflaton via yukawa interaction @xmath148 . for successful inflation the running of @xmath127 must be sufficiently strong to generate a local extremum of the potential @xmath153 for some nonvanishing field value , which we call @xmath154
. the inflaton potential will obviously be flat near @xmath154 , so that inflation usually occurs at field values not very far from @xmath154 .
we therefore expand @xmath155 around @xmath156 .
the potential we work with thus reads : @xmath157 here @xmath158 is given by the @xmath145function of eq.([beta ] ) , and @xmath159 is given by the scale dependence of the parameters appearing in eq.([beta ] ) .
in contrast to earlier analyses of this model @xcite , we include the @xmath160 term in the potential .
this is a two loop correction , but it can be computed by `` iterating '' the one loop correction .. they are thus formally included in our coefficient @xmath161 . ]
since the coefficient @xmath162 of this term is of fourth order in couplings , one will naturally expect @xmath163 . however , this need not be true if @xmath164 `` happens '' to be suppressed by a cancellation in eq.([beta ] ) . including
the second correction to @xmath155 seems natural given that we also expanded the running of the spectral index to second ( quadratic ) order .
recall that we had defined @xmath154 to be a local extremum of @xmath153 , i.e. @xmath165 .
this implies @xcite @xmath166 ; this relation is not affected by the two loop correction @xmath167 . in order to calculate the spectral parameters @xmath168 and @xmath79 defined in eqs.([runpar ] ) , we need the first four slow roll parameters , defined as @xcite : and @xmath169 are purely by convention ; in particular , @xmath170 could be negative . ]
@xmath171 where primes denote derivatives with respect to @xmath124 .
all these parameters are in general scale dependent , i.e. they have to be evaluated at the value of @xmath124 that the inflaton field had when the scale @xmath56 crossed out of the horizon .
the spectral parameters are related to these slow roll parameters by @xcite : @xmath172 slow
roll inflation requires @xmath173 . combining eqs.([pot3 ] ) and ( [ slow - roll ] )
we see that we need @xmath174 , where we have introduced the short hand notation @xmath175 in other words , the inflaton potential has to be dominated by the constant term , as noted earlier .
eqs.([slow - roll ] ) then imply two strong inequalities between ( combinations of ) slow roll parameters : @xmath176 is essentially determined by @xmath177 .
similarly , both relations together imply that @xmath78 is basically fixed by @xmath170 , while only the last two terms in the expression for @xmath79 are relevant ; these two terms are generically of similar order of magnitude .
finally , the factors of @xmath178 appearing in the denominators of eqs.([slow - roll ] ) can be replaced by @xmath179 .
the spectral parameters are thus given by : @xmath180\ , ; \nonumber \\ % \alpha_s & = & -2 \left ( \frac { c m_{\rm p}^2}{v_0 } \right)^2 l \left [ 1 + \frac{g } { 2c } \left ( 2l + 3 \right ) \right ] \left [ 1 + \frac { g } { 2c } \left ( l + 1 \right ) \right ] \ , ; \\ % \beta_s & = & 2 \left ( \frac { c m_{\rm p}^2}{v_0 } \right)^3 l \left [ 1 + \frac { g}{2c } \left ( l+1 \right ) \right ] \left [ 1 + \frac { g}{2c } \left ( 3l+2 \right ) + \frac { g^2 } { 2c^2 } \left ( 3 l^2 + 5l + \frac{3}{2 } \right ) \right ] \nonumber \ , , \end{aligned}\ ] ] where @xmath181 has been defined in eq.([l ] ) .
clearly the spectral index is not scale invariant unless @xmath161 and @xmath162 are very close to zero .
note that @xmath179 appears in eqs.([nab - slo ] ) only in the dimensionless combination @xmath182 , while @xmath124 only appears via @xmath181 , i.e. only the ratio @xmath183 appears in these equations .
in contrast , the absolute normalization of the power spectrum is given by ( in slow
roll approximation ) @xmath184 this normalization is usually quoted at the `` cobe scale '' @xmath66 mpc@xmath67 , where @xmath185 @xcite . applying our potential ( [ pot3 ] ) to eq.([power ] ) , replacing @xmath178 by @xmath179 in the numerator
, we see that @xmath186 not only depends on @xmath187 and @xmath181 , but also on the ratio @xmath188 . we can thus always find parameters that give the correct normalization of the power spectrum , for all possible combinations of the spectral parameters .
we want to find out whether the potential ( [ pot3 ] ) can accommodate sufficient running of @xmath59 to allow pbh formation .
there are strong observational constraints on @xmath59 and @xmath78 .
it is therefore preferable to use these physical quantities directly as inputs , rather than the model parameters @xmath189 and @xmath190 . to this end
we rewrite the first eq.([nab - slo ] ) as : @xmath191 inserting this into the second eq.([nab - slo ] ) gives : @xmath192 \left [ 1 + \frac { g}{2c } \left ( l+1 \right ) \right ] } { \left [ l + 1 + \frac { g } { 2c } \left ( l^2 + 3l + 1 \right ) \right]^2 } \ , .\ ] ] we thus see that the running of the spectral index is `` generically '' of order @xmath193 ; similarly , the running of the running can easily be seen to be @xmath194 .
this is true in nearly all inflationary scenarios that have a scale dependent spectral index .
eq.([sol2 ] ) can be solved for @xmath190 .
bringing the denominator to the left
hand side leads to a quadratic equation , which has two solutions .
they can be written as : @xmath195 + r_s^2 \left ( 0.5l + 1 \right)^2 } } { \left ( l^2 + 3l + 1 \right)^2 + r_s l \left ( 2l + 3 \right ) \left ( l+1 \right ) } \,,\ ] ] where we have introduced the quantity @xmath196 since @xmath162 and @xmath161 are real quantities , eq.([sol3 ] ) only makes sense if the argument of the square root is non negative .
note that the coefficients multiplying @xmath197 and @xmath198 inside the square root are both non negative .
this means that the model can in principle accommodate any non negative value of @xmath197 .
however , small negative values of @xmath197 can not be realized .
it is easy to see that the constraint on @xmath197 is weakest for @xmath199 .
the argument of the square root is then positive if @xmath200 , which implies either @xmath201 or @xmath202 .
recalling the definition ( [ rs ] ) we are thus led to the conclusion @xmath203 this bound should hold on all scales , as long as the potential is described by eq.([pot3 ] ) .
note that it is identical to the bound found in ref.@xcite , i.e. it is not affected by adding the term @xmath204 to the inflaton potential .
this is somewhat disappointing , since recent data indicate that @xmath78 is negative at cmb scales .
even the generalized version of the running mass model therefore can not reproduce the current @xmath205 range of @xmath78 .
however , at the @xmath206 level significantly positive @xmath78 values are still allowed .
let us therefore continue with our analysis , and search for combinations of parameters within the current @xmath206 range that might lead to significant pbh formation . using eqs.([sol1 ] ) and ( [ sol3 ] ) we can use @xmath207 and @xmath208 as input parameters in the last eq.([nab - slo ] ) to evaluate @xmath209 .
this can then be inserted into eq.([nofr ] ) to see how large the density perturbations at potential pbh scales are .
this numerical analysis is most easily performed at the `` pivot scale '' , where the errors on @xmath59 and @xmath78 are essentially uncorrelated ; at @xmath57values above ( below ) this scale , @xmath59 and @xmath78 are correlated ( anticorrelated ) @xcite . the pivot scale for the
`` wmap7+bao+@xmath110 '' data set we are using is @xmath210 mpc@xmath67 @xcite .
values given in ref.@xcite .
the difference between these two values is not important for our numerical analysis . ] at this scale , observations give @xcite : @xmath211 we saw above that requiring the correct normalization of the power spectrum at cmb scales does not impose any constraint on the spectral parameters .
however , the model also has to satisfy several consistency conditions . to begin with
, it should provide a sufficient amount of inflation . in the slow
roll approximation , the number of e folds of inflation following from the potential ( [ pot3 ] ) is given by : @xmath212 } \nonumber \\ & = & - \frac { 2 } { \tilde c \left ( 1 + \frac { g}{2c } \right ) } \left [ \ln \frac { l } { l_0 } - \ln \frac { 1 + \frac { g}{2c } ( l+1 ) } { 1 + \frac { g}{2c } ( l_0 + 1 ) } \right]\,,\end{aligned}\ ] ] where we have introduced the dimensionless quantity @xmath213 recall that it can be traded for @xmath214 using eq.([sol1 ] ) .
moreover , @xmath215 can be related to the scale @xmath56 through @xmath216 this can be inverted to give @xmath217 } \,,\ ] ] where we have introduced @xmath218 the problem is that the denominator in eq.([lofk ] ) vanishes for some finite value of @xmath56 .
this defines an extremal value of @xmath219 : @xmath220\,.\ ] ] a negative value of @xmath221 is generally not problematic , since only a few e folds of inflation have to have occurred before our pivot scale @xmath222 crossed out of the horizon .
however , a small positive value of @xmath223 would imply insufficient amount of inflation after the scale @xmath222 crossed the horizon . in our numerical work
we therefore exclude scenarios with @xmath224 , i.e. we ( rather conservatively ) demand that at least 50 e folds of inflation can occur after @xmath222 crossed out of the horizon .
a second consistency condition we impose is that @xmath225 should not become too large . specifically , we require @xmath226 for all scales between @xmath222 and the pbh scale
. for ( much ) larger values of @xmath225 our potential ( [ pot3 ] ) may no longer be appropriate , i.e. higher powers of @xmath181 may need to be included .
note that eq.([lofk ] ) allows to compute the effective spectral index @xmath227 _ exactly _ : @xmath228 \frac { l(k ) + 1 + \frac { g}{2c } \left [ l(k)^2 + 3 l(k ) + 1 \right ] } { l_0 + 1 + \frac{g}{2c } \left ( l_0 ^ 2 + 3 l_0 + 1 \right ) } \,.\ ] ] this in turn allows an _ exact _ ( numerical ) calculation of the spectral index @xmath229 : @xmath230 d \
, \ln k^\prime \ , .\ ] ] in our numerical scans of parameter space we noticed that frequently the exact value for @xmath229 at pbh scales differs significantly from the values predicted by the expansion of eq.([nofr ] ) ; similar statements apply to @xmath227 .
this is not very surprising , given that @xmath231 for our value of the pivot scale and @xmath105 g .
in fact , we noticed that even if @xmath113 and @xmath209 are both positive , @xmath227 may not grow monotonically with increasing @xmath56 . in some cases
@xmath227 computed according to eq.([nsofk ] ) even becomes quite large at values of @xmath56 some 5 or 10 e folds below the pbh scale .
this is problematic , since our calculation is based on the slow
roll approximation , which no longer works if @xmath232 becomes too large .
we therefore demanded @xmath233 for all scales up to the pbh scale ; the first eq.([nab ] ) shows that this corresponds to @xmath234 .
this last requirement turns out to be the most constraining one when looking for combinations of parameters that give large @xmath235 .
vs. @xmath113 . here the model parameters @xmath236 and @xmath237 are scanned randomly , with flat probability distribution functions . ]
we are now ready to present some numerical results .
we begin in figure [ fig : fig2 ] , which shows a scatter plot of the spectral parameters @xmath113 and @xmath209 , which has been obtained by randomly choosing model parameters @xmath238 [ defined in eq.([ct ] ) ] , @xmath190 and @xmath237 in the ranges . ]
we require that @xmath112 and @xmath113 lie within their @xmath206 ranges , and impose the consistency conditions discussed above .
the plot shows a very strong correlation between @xmath79 and @xmath78 : if the latter is negative or small , the former is also small in magnitude
. moreover , there are few points at large @xmath78 , and even there most allowed combinations of parameters lead to very small @xmath79 .
the accumulation of points at small @xmath78 can be understood from our earlier result ( [ sol2 ] ) , which showed that @xmath78 is naturally of order @xmath240 within @xmath241 .
moreover , @xmath79 is naturally of order @xmath242 . on the other hand , for @xmath78 values close to the upper end of the current @xmath206 range
, we do find some scenarios where @xmath79 is sufficiently large to allow the formation of @xmath6 g pbhs .
we also explored the correlation between @xmath112 and @xmath113 ( not shown ) . here
the only notable feature is the lower bound ( [ asmin ] ) on @xmath113 ; values of @xmath113 up to ( and well beyond ) its observational upper bound can be realized in this model for any value of @xmath112 within the presently allowed range .
similarly , we do not find any correlation between @xmath112 and @xmath209 .
this lack of correlation can be explained through the denominator in eq.([sol2 ] ) , which also appears ( to the third power ) in the expression for @xmath79 once eq.([sol1 ] ) has been used to trade @xmath161 for @xmath232 : this denominator can be made small through a cancellation , allowing sizable @xmath78 even if @xmath59 is very close to 1 . since the same denominator appears ( albeit with different power ) in the expressions for @xmath78 and @xmath79 , it does not destroy the correlation between these two quantities discussed in the previous paragraph .
( in units of @xmath243 ; double dotted [ green ] curve ) , @xmath244 ( dotted , blue ) , the effective spectral index @xmath232 at the pbh scale ( dot dashed , red ) and the spectral index @xmath245 at the pbh scale ( solid , black ) are shown as functions of @xmath246 , for @xmath247 and @xmath248 . if both solutions in eq.([sol3 ] ) for @xmath190 are acceptable , we have taken the one giving larger @xmath73 at the pbh scale . ]
figure [ fig : fig2 ] indicates that large values of @xmath73 at the pbh scale can be achieved only if @xmath78 at the cmb scale is positive and not too small . in figure
[ fig : fig3 ] we therefore explore the dependence of the potential parameters , and of the spectral parameters at the pbh scale on @xmath237 , for @xmath249 .
note that varying @xmath237 also changes the parameters @xmath161 and @xmath162 ( or @xmath190 ) , see eqs.([sol1 ] ) and ( [ sol3 ] ) .
the latter in general has two solutions ; however , for most values of @xmath237 , only one of them leads to sufficient inflation while keeping @xmath250 ; if both solutions are allowed , we take the one giving a larger spectral index at the pbh scale , taken to be @xmath251 mpc@xmath67 corresponding to @xmath105 g. note that @xmath59 and @xmath73 at the pbh scale are calculated exactly , using eqs.([nab ] ) and ( [ nofk ] ) .
we find that the expansion ( [ nofr ] ) is frequently very unreliable , e.g. giving the wrong sign for @xmath245 at the pbh scale for @xmath252 .
figure [ fig : fig3 ] shows that @xmath238 is usually well below 1 , as expected from the fact that @xmath253 , see the first eq.([nab - slo ] ) .
moreover , in most of the parameter space eq.([sol3 ] ) implies @xmath254 ; recall that this is also expected , since @xmath162 is a two loop term .
we find @xmath255 only if @xmath164 is small .
in particular , the poles in @xmath244 shown in figure [ fig : fig3 ] occur only where @xmath161 vanishes ; note that the spectral parameters remain smooth across these `` poles '' .
there are a couple of real discontinuities in figure [ fig : fig3 ] , where the curves switch between the two solutions of eq.([sol3 ] ) .
the first occurs at @xmath256 . for smaller values of @xmath237 ,
the solution giving the smaller @xmath257 violates our slow
roll condition @xmath258 at scales close to the pbh scale . for larger @xmath237
this condition is satisfied . just above the discontinuity , where @xmath59 is close to 2 at the pbh scale , we find the largest spectral index at the pbh scale , which is close to 0.47 .
recall from figure [ fig : fig1 ] that this will generate sufficiently large density perturbations to allow the formation of pbhs with @xmath105 g. however , the formation of pbhs with this mass is possible only for a narrow range of @xmath237 , roughly @xmath259 . at @xmath260
, @xmath161 goes through zero , giving a pole in @xmath190 as discussed above .
then , at @xmath261 , the second discontinuity occurs . here
the curves switch between the two solutions of eq.([sol3 ] ) simply because the second solution gives a larger spectral index at the pbh scale .
right at the discontinuity both solutions give the same spectral index , i.e. the curve depicting @xmath262 remains continuous ; however , @xmath238 jumps from about @xmath263 to @xmath264 .
the effective spectral index @xmath59 at the pbh scale also shows a small discontinuity .
recall from our discussion of eq.([nsofn ] ) that @xmath59 will generally be larger than @xmath73 at the pbh scale , but the difference between the two depends on the model parameters .
for very small values of @xmath265 , @xmath238 becomes very large ; this region of parameter space is therefore somewhat pathological . for sizably positive @xmath237 ,
@xmath73 at the pbh scale increases slowly with increasing @xmath237 , while @xmath266 and @xmath257 both decrease
. however , the spectral index at the pbh scale remains below the critical value for the formation of long lived pbhs .
note that @xmath181 always maintains its sign during inflation , since @xmath199 corresponds to a stationary point of the potential , which the ( classical ) inflaton trajectory can not cross . for most of the parameter space shown in figure [ fig : fig3 ] ,
@xmath225 decreases during inflation . if @xmath267 decreasing @xmath225 corresponds to @xmath268 , i.e. the inflaton rolls towards a minimum of the potential at @xmath269 . for @xmath270
we instead have @xmath271 , i.e. the inflaton rolls away from a maximum of the potential .
in fact , this latter situation also describes the branch of figure [ fig : fig3 ] giving the largest spectral index at the pbh scale ; since here @xmath267 , @xmath225 increases during inflation on this branch .
this is illustrated in figures [ fig : fig4 ] , which show the ( rescaled ) inflaton potential as well as the effective spectral index as function of either the inflaton field ( left frame ) or of the scale @xmath56 ( right frame ) . note that all quantities shown here are dimensionless , and are determined uniquely by the dimensionless parameters @xmath238 defined in eq.([ct ] ) , @xmath190 and @xmath237 .
this leaves two dimensionful quantities undetermined , e.g. @xmath179 and @xmath154 ; one combination of these quantities can be fixed via the normalization of the cmb power spectrum , leaving one parameter undetermined ( and irrelevant for our discussion ) .
the left frame shows that the ( inverse ) scale @xmath56 first increases quickly as @xmath124 rolls down from its initial value @xmath272 .
this means that @xmath124 initially moves rather slowly , as can also be seen in the right frame . since @xmath273 , the effective spectral index increases with increasing @xmath56
the right frame shows that this evolution is quite nonlinear , although for @xmath274 , @xmath227 is to good approximation a parabolic function of @xmath275 .
however , for even smaller scales , i.e. larger @xmath56 , the rate of growth of @xmath59 decreases again , such that @xmath232 reaches a value very close to @xmath81 at the scale @xmath276 relevant for the formation of pbhs with @xmath105 g. recall that we only allow solutions where @xmath233 for the entire range of @xmath56 considered ; figures [ fig : fig4 ] therefore illustrate our earlier statement that this constraint limits the size of the spectral index at pbh scales .
the left frame of figure [ fig : fig4 ] shows that the inflaton potential as written becomes unbounded from below for @xmath277 .
this can be cured by introducing a quartic ( or higher ) term in the inflaton potential ; the coefficient of this term should be chosen sufficiently small not to affect the discussion at the values of @xmath124 of interest to us .
note also that this pathology of our inflaton potential is not visible in the right frame , since assuming @xmath278 at @xmath279 , the inflaton field can never have been larger than @xmath154 : as noted above , it can not have moved across the maximum of the potential .
that is consistent with the constraints we impose , as function of @xmath113 .
the other curves show the corresponding model parameters : @xmath238 ( double - dotted , green ) , multiplied with @xmath280 for ease of presentation ; @xmath281 ( dotted , blue ) ; and @xmath282 ( dot dashed , red ) . @xmath112 is fixed at 0.964 , but the bound on @xmath262 is almost independent of this choice . ]
figure [ fig : fig2 ] indicated a strong dependence of the maximal spectral index at pbh scales on @xmath113 .
this is confirmed by figure [ fig : fig5 ] , which shows the maximal possible @xmath235 consistent with our constraints as function of @xmath113 , as well as the corresponding values of the parameters @xmath283 and @xmath237 .
we saw in the discussion of eq.([sol3 ] ) that @xmath284 is only allowed for a narrow range of @xmath237 . in this very constrained corner of parameter space
, @xmath235 remains less than @xmath81 , although the effective spectral index @xmath285 can exceed @xmath81 for @xmath286 ; recall that for the given choice @xmath247 , solutions only exist if @xmath287 , see eq.([asmin ] ) . for @xmath288
the optimal set of parameters lies well inside the region of parameter space delineated by our constraints .
@xmath285 therefore grows very fast with increasing @xmath113 . for these very small values of @xmath113 , the largest spectral index at pbh scales
is always found for @xmath289 , which implies that the second derivative of the inflaton potential also vanishes at @xmath156 , i.e. @xmath154 corresponds to a saddle point , rather than an extremum , of the potential . for slightly larger values of @xmath113
the choice @xmath289 allows less than 50 e folds of inflation after @xmath222 exited the horizon .
requiring at least 50 e folds of inflation therefore leads to a kink in the curve for @xmath235 .
the optimal allowed parameter set now has considerable smaller @xmath257 , but larger @xmath266 .
the curve for the maximal @xmath235 shows a second kink at @xmath290 . to the right of this point
the most important constraint is our requirement that @xmath258 at all scales up to @xmath291 , as discussed in connection with figures [ fig : fig3 ] and [ fig : fig4 ] .
note that for @xmath292 , the optimal parameter choice leads to @xmath59 reaching its maximum at some intermediate @xmath56 close to , but smaller than , @xmath293 .
this leads to a further flattening of the increase of @xmath235 .
we nevertheless see that for values of @xmath113 close to the upper end of the @xmath241 range specified in ( [ range ] ) the spectral index at the scale relevant for the formation of @xmath6 g pbhs can be well above the minimum for pbh formation found in section 2 .
figure [ fig : fig1 ] then implies that the formation of considerably heavier pbhs might be possible in running mass inflation . however , larger pbh masses correspond to smaller @xmath291 , see eq.([r ] ) .
this in turn allows for less running of the spectral index . in order to check whether even heavier pbhs might be formed during the slow
roll phase of running mass inflation , one therefore has to re optimize the parameters for different choices of @xmath293 . that allows the formation of primordial black holes of a given mass as a result of density perturbations produced during the slow
roll phase of inflation . ]
the result of such an analysis is shown in figure [ fig : fig6 ] .
we see that in the given model , the formation of pbhs that are sufficiently massive , and hence long lived , to be cdm candidates could only have been triggered during the slow
roll phase of inflation if @xmath113 is more than one standard deviation above its central value .
if @xmath113 is at the upper end of its present @xmath241 range , pbhs with mass up to @xmath294 g could have formed as result of density perturbations created at the end of slow
roll inflation .
these results are again insensitive to the value of @xmath112 .
[ [ section-3 ] ] in this paper we have investigated the formation of primordial black holes in the radiation dominated era just after inflation .
we have focused on density perturbations originating from the slow
roll phase of inflation . in section 2
we reviewed the press
schechter type formalism for pbh formation .
we found that the formation of pbhs with mass larger than @xmath6 g , which could form ( part of ) the cold dark matter in the universe , the spectral index at scale @xmath293 should be at least @xmath295 , even for the lower value of @xmath296 for the threshold @xmath297 . this value is higher that the value @xmath298 found in @xcite because here we have assumed that the mass of the collapsed region to form pbh is only 20% of the entire energy density inside the particle horizon .
this spectral index is much above the value measured in the cmb .
pbh formation therefore requires significant positive running of the spectral index when @xmath56 is increased .
in contrast , current observations favor a negative first derivative of the spectral index . for the central value @xmath247 , where @xmath222 is the cmb pivot scale , the first derivative @xmath113 would need to exceed @xmath299 if it alone were responsible for the required increase of the spectral index ;
this is outside the current @xmath300 range for this quantity .
however , the second derivative ( the `` running of the running '' ) is currently only very weakly constrained .
we showed in a model
independent analysis that this easily allows values of @xmath235 large enough for pbh formation , even if the first derivative of the spectral index is negative at cmb scales . in section 3
we applied this formalism to the `` running mass '' model of inflation , which had previously been shown to allow sufficiently large positive running of the spectral index to yield pbh formation .
in contrast to previous analyses , we included a term quadratic in the logarithm of the field , i.e. we included the `` running of the running '' of the inflaton mass along with the `` running of the running '' of the spectral index .
we showed that this model can accommodate a sizably positive second derivative of the spectral index at pbh scales .
however , this is only possible if the first derivative is also positive and sizable .
in fact , like most inflationary scenarios with a smooth potential @xcite the model does not permit large negative running of the spectral index at cmb scales .
current data prefere negative running of the spectral index at cmb scales , but are consistent with sufficiently large positive running at the @xmath206 level .
moreover , we saw that a quadratic ( in @xmath301 ) extrapolation of the spectral index to pbh scales is not reliable , and therefore computed the spectral index exactly . imposing several consistency conditions
, we found that density perturbations that are sufficiently large to trigger pbh formation only occur for a very narrow region of parameter space . among other things ,
the signs of the parameters of the inflaton potential must be chosen such that the potential has a local maximum , and the initial value of the field must be slightly ( by typically less than one e fold ) below this maximum .
we emphasize that obtaining sufficiently large density perturbations at small scales is only a necessary condition for successful pbh formation
. one major challenge of this model is that for parameters allowing pbh formation the spectral index keep increasing at yet smaller scales .
parameters that lead to the formation of many pbhs with mass around @xmath6 g , which could form the dark matter in the universe , would predict the over production of unstable pbhs , in conflict with data e.g. from the non observation of black hole evaporation and ( for yet smaller masses ) big bang nucleosynthesis .
this problem seems quite generic for this mechanism .
one way to solve it might be to abruptly cut off inflation just after the scales relevant for the formation of the desired pbhs leave the horizon , which could e.g. be achieved by triggering the waterfall field in hybrid inflation .
however , this is somewhat in conflict with the use of the slow
roll formalism of structure formation , which is usually assumed to require a few more e folds of inflation after the scales of interest left the horizon.moreover , a sharpe end of inflation means that the visible universe only inflated by about @xmath302 e folds ; this solves the horizlatness problems only if the reheat temperature was rather low @xcite .
we can therefore not state with confidence that formation of pbhs as dark matter candidates is possible in running
mass inflation ; all we can say is that certain necessary conditions can be satisfied . even that is possible only for a very limited range of parameters .
this also means that constraints from the over production of pbhs only rule out a small fraction of the otherwise allowed parameter space of this model .
this work is supported in part by the dfg transregio tr33 `` the dark universe '' . ee thanks the bonn cologne graduate school for support , and the kias school of physics for hospitality .
md thanks the particle theory group of the university of hawaii at manoa for hospitality while this work was completed .
a. s. josan , a. m. green and k. a. malik , _ phys .
_ * d79 * ( 2009 ) , 103520 [ arxiv : 0903.3184 ] ; b. carr , k. kohri , y. sendouda and j. yokoyama , _ phys . rev . _
* d81 * ( 2010 ) 104019 [ arxiv : astro - ph/0912.5297 ] .
s. m. leach , i. j. grivell and a. r. liddle , _ phys .
_ * d62 * ( 2000 ) 043516 [ arxiv : astro - ph/0004296 ] ; k. kohri , d. h. lyth and a. melchiorri , _ jcap _ * 0804 * ( 2008 ) 038 [ arxiv : hep - ph/0711.5006 ] ; l. alabidi and k. kohri , _ phys .
* d80 * ( 2009 ) 063511 2009 [ arxiv : 0906.1398 ] ; e. bugaev , p. klimai , _ phys .
_ * d79 * ( 2009 ) 103511 [ arxiv : 0812.4247 ] .
j. h. macgibbon , _ nature _ * 320 * ( 1987 ) 308 ; b. j. carr , j. h. gilbert and j. e. lidsey , _ phys .
_ * d50 * ( 1994 ) 4853 [ arxiv : astro - ph/9405027 ] ; j. d. barrow , e. j. copeland and a. r. liddle , _ phys .
_ * d46 * ( 1992 ) 645 ; a. bonanno and m. reuter , _ phys .
rev . _ * d62 * ( 2000 ) 043008 [ arxiv : hep - th/0002196 ] . a. r. liddle and d. h. lyth , _
cosmological inflation and large - scale structure _ , cambridge university press , 2000 ; d. h. lyth and a. r. liddle , _ the primordial density perturbation _ , cambridge university press , 2009 ; q. huang , _ jcap _ * 0611 * ( 2006 ) 004 [ arxiv : astro - ph/0610389 ] .
d. h. lyth , k. a. malik , m. sasaki and i. zaballa , _ jcap _ * 0601 * ( 2006 ) 011 [ arxiv : astro - ph/0510647 ] ; i. zaballa , a. m. green , k. a. malik and m. sasaki , _ jcap _ * 0703 * ( 2007 ) 010 [ arxiv : astro - ph/0612379 ] . | we revisit the question whether the running mass inflation model allows the formation of primordial black holes ( pbhs ) that are sufficiently long lived to serve as candidates for dark matter .
we incorporate recent cosmological data , including the wmap 7year results .
moreover , we include `` the running of the running '' of the spectral index of the power spectrum , as well as the renormalization group `` running of the running '' of the inflaton mass term .
our analysis indicates that formation of sufficiently heavy , and hence long lived , pbhs still remains possible in this scenario . as a by product
, we show that the additional term in the inflaton potential still does not allow significant negative running of the spectral index .
and _ @xmath0physikalisches institut and bethe center for theoretical physics , universitt bonn , + nussallee 12 , 53115 bonn , germany + and + @xmath1school of physics , kias , seoul 130722 , korea _ |
the studies of low - dimensional magnets have presented a topical item for many years .
the investigations include both 1d and 2d magnets , as well as bilayers , thin films and multilayers .
many different theoretical methods have been employed ; in some cases [ 16 ] the exact solutions are available . as far as approximate methods are concerned
, one should mention the green function ( gf ) [ 79 ] and spectral density [ 10 ] methods , renormalization group ( rg ) approach [ 11 , 12 ] , including mean - field renormalization group ( mfrg ) [ 13 , 14 ] and effective - field renormalization group ( efrg ) [ 15 ] , spin - wave ( sw ) techniques [ 1618 ] , high temperature series expansion ( htse ) [ 19 ] and monte carlo ( mc ) simulations [ 20 , 21 ] .
many other approaches like coherent anomaly method ( cam ) together with transfer matrix technique [ 22 ] and effective field methods ( efm ) with correlations [ 23 ] should also be mentioned .
recently , cluster variational method in the pair approximation ( pa ) has been adopted for studies of the bilayer [ 24 ] and bi - multilayer [ 25 ] systems .
the aim of the present paper is to study the ferromagnetic ising model with spin @xmath0 on the planar triangular ( pt ) lattice with selective dilution . by the selective dilution
we mean the dilution of only one sublattice , whereas the pt magnet can be , in general , decomposed into three interpenetrating sublattices .
regarding antiferromagnetism and the problem of frustration , such selectively diluted model has been considered by kaya and berker [ 26 ] .
to the best of our knowledge , as far as ferromagnetism is concerned , the model has not been studied yet .
the model is interesting from the theoretical point of view , for by changing the selective dilution parameter we are able to pass continuously from the ideal pt lattice ( without dilution ) to the honeycomb lattice , where one sublattice is completely empty . on the other hand , for those two cases the exact solutions for the curie temperatures do exist [ 2 ] . in this paper
we will concentrate on the curie temperature calculations for arbitrary concentration of magnetic atoms in the selectively diluted sublattice .
thus , we consider an intermediate situation between those two limiting cases , which were examined exactly by wannier [ 2 ] . in the next section , the outline of the theoretical methods will be given , and the respective formulas for the phase transition temperatures will be presented . with the help of numerical calculations we are able to compare the results of several theoretical approaches , namely the molecular field approximation ( mfa ) , effective field theory ( eft ) , pair approximation ( pa ) method , as well as the exact calculation for finite clusters ( ecfc ) .
the results will be presented in the plots and discussed .
let us consider the ferromagnetic ising model with spin @xmath0 on the pt lattice with selective dilution .
the diluted lattice is illustrated in fig.1 . following kaya and berker
[ 26 ] we keep decomposition of the pt lattice into three interpenetrating sublattices @xmath1 , @xmath2 and @xmath3 , of which , for the ferromagnetic case the sublattices @xmath1 and @xmath2 will be equivalent . the sublattice @xmath3 is distinguished in the system by the random dilution of the spins .
if we denote by @xmath4 the concentration of spins on @xmath3-sublattice , the case of @xmath5 corresponds to the honeycomb lattice , whereas @xmath6 stands for the non - diluted triangular lattice .
the hamiltonian of the system is in the form of : @xmath7 where @xmath8 is the exchange interaction coupling , and @xmath9 means the summation extending over nearest neighbour pairs of spins from sublattice @xmath10 and @xmath11 .
@xmath12 are quenched , uncorrelated random variables chosen to be equal 1 with probability @xmath4 when the site @xmath13 is ocupied by a magnetic atom and 0 with probability @xmath14 otherwise . ,
@xmath2 and @xmath3 . the sublattice @xmath3 is randomly diluted .
the thick grey line surrounds a cluster containing @xmath15 hexagons , used for exact calculations . ] in this simplest approach the sublattice magnetizations @xmath16 , @xmath17 and @xmath18 are described by three coupled mfa equations . in the vicinity of the curie temperature the equations can be linearized and presented as follows : @xmath19 where @xmath20 , and @xmath21 denotes the curie temperature .
it is convenient to introduce the variable @xmath22 which is a dimensionless curie temperature .
then , assuming symmetry condition for the ferromagnet @xmath23 , and demanding that the determinant of eqs.(2 ) must be zero , we obtain the equation for the curie temperature in mfa : @xmath24 the physical solution is then of the form : @xmath25 and is straightforward for numerical calculation . by the eft we mean the method proposed by honmura and kaneyoshi [ 27 ] , which takes into account autocorrelations but neglects the spin - pair correlations . among its many applications ,
the method has recently been applied for the triangular lattice with uniform dilution [ 28 ] .
it is based on the exact callen - suzuki identity of the form : @xmath26 where @xmath27 , if @xmath28 and @xmath29 denotes a lattice site being nearest neighbour of the site @xmath30 . applying the differential operator method [ 27 ] , together with the decoupling procedure for the mean value of multi - spin products ,
the local magnetizations can be calculated .
the coupled equations for the sublattice magnetizations have polynomial form and can be linearized near the continuous phase transition points . for the system in question , with the general sublattice magnetizations @xmath31 , @xmath32 and @xmath33
, the curie temperature can be found from the determinant equation : @xmath34 where @xmath35 in eq.(7 ) the temperature - dependent coefficients are given in the following form : @xmath36 where @xmath37 is the differential operator .
the equation([eq5 ] ) can be solved numerically only .
the pa is one of the cluster variational methods and takes into account the nearest - neighbour correlations .
being more accurate than mfa and eft it enables calculation of the gibbs energy , and hence all thermodynamic properties .
contrary to mfa , in the pa method the local variational parameters ( molecular fields acting on a pair ) are no longer simply proportional to local magnetizations .
a set of linearized equations for these parameters near the curie temperature takes a form : @xmath38 the variational parameters have the following meaning : + @xmath39 is the field acting on a spin on the sublattice @xmath1 or @xmath2 and originating from spins on the sublattices @xmath2 or @xmath1 , respectively ; + @xmath40 is the field acting on the spin on the sublattice @xmath3 and originating from the sublattices @xmath1 or @xmath2 ; + @xmath41 is the field acting on the spin on the sublattice @xmath1 or @xmath2 and originating from the sublattice @xmath3 .
the temperature - dependent coefficients have the following form : @xmath42 where @xmath43 by setting the determinant of eqs.(9 ) equal to zero the curie temperature can be found .
the final result can be presented in the form of the algebraic equation : @xmath44 where @xmath45 is related to the curie temperature , @xmath46 , by the formula ( 11 ) .
exact numerical diagonalization for finite systems is nowadays a powerful tool for the studies of magnetic properties [ 29 ] .
the accuracy of this method improves with the increase of the cluster size , but simultaneously rapidly growing number of states , which should be taken into account , results in a corresponding huge increase of the calculation time .
this requires increasingly powerful computers .
however , in the case of spin systems with ising interactions , all the system states and their energies can be listed explicitly , without resorting to diagonalization of the hamiltonian .
hence , we call this approach exact calculation for finite clusters .
the method bears some resemblance to monte carlo calculations , however , it uses all the system states to study its thermodynamics within canonical ensemble .
the method is sensitive to the shape of a cluster and selection of the boundary conditions .
it is known from the literature devoted to monte carlo simulations that the selection of periodic boundary conditions is evaluated as superior to other choices for planar lattices [ 30,31 ] .
in particular , it guarantees the same number of nearest - neighbours for the atoms on the boundary and inside the cluster . in our calculations presented here
, we based on a cluster consisting of @xmath47 hexagons , and the boundary conditions were chosen as periodic .
such a cluster is illustrated in fig.[fig1 ] , where it is surrounded by the gray thick line .
the maximum number of spins ( for @xmath6 , when all sites on the sublattice @xmath3 were occupied ) amounted to 36 , while for @xmath5 it was equal to 24 .
the curie temperatures were determined from the maxima of the specific heat curves for various concentrations @xmath4 .
the condition for the maximum can be found from the exact thermodynamic formula : @xmath48k_{\rm b}t_{\rm max},\ ] ] where the mean values of the energy powers @xmath49 are calculated numerically with the boltzmann distribution ( taken at the temperature @xmath50 ) over all possible states in the cluster . for our purpose , in eq.(13 ) @xmath50 for a finite cluster is assumed to estimate @xmath21 , which is a common approach used in monte carlo studies , e.g. [ 32,33 ] . according to scaling relations
, @xmath50 is expected to converge to @xmath21 in the limit of an infinite system size [ 33,34 ] . in the case of ecfc
, the system size is severely limited by the computational resources and thus extrapolation to infinite system size would be not trustworthy , due to small - size corrections to scaling . therefore , we present directly the obtained values of @xmath50 for the largest system studied , i.e. the 3@xmath514 cluster and for comparison we provide also the numbers for smaller 3@xmath513 cluster .
the numerical results are presented in the next section .
the curie temperatures for selectively diluted pt ferromagnet have been calculated based on the approximations presented in the previous section .
the results of @xmath52 vs. concentration @xmath4 are illustrated in fig.2 . in the same figure two exact wannier results [ 2 ] are shown , i.e. , for @xmath5 ( honeycomb lattice ) and for @xmath6 ( triangular lattice ) .
for the pure honeycomb lattice we obtained the curie temperature values @xmath52 equal to : 3/4 ( mfa ) ; 0.5259 ( eft ) ; @xmath530.4551 ( pa ) ; 0.4203 ( ecfc for 3@xmath513 cluster ) and 0.4128 ( ecfc for 3@xmath514 cluster ) .
these results can be compared with the exact wannier solution @xmath540.3797 . on the other hand , for the pure triangular lattice
the curie temperatures calculated in various approaches are : @xmath52=3/2 ( mfa ) ; 1.2683 ( eft ) ; @xmath551.2332 ( pa ) ; 0.9602 ( ecfc for 3@xmath513 cluster ) and 0.9520 ( ecfc for 3@xmath514 cluster ) . the exact wannier result in this case is @xmath560.9102 .
4 cluster .
two exact wannier results [ 2 ] for @xmath5 and @xmath6 are also indicated . ]
the presented results are obtained by the ecfc method for 3@xmath514 cluster . ]
the curie temperature of the pure triangular lattice is higher than that of the honeycomb one , since the coordination number of the former doubles that of the latter .
it is seen in fig.2 that in each method the curie temperature changes continuously with @xmath4 ; however , the change is not linear , not even in the mfa .
fig.2 illustrates the accuracy of the methods described in the previous section .
as far as the analytical methods are concerned , we see that mfa is the least accurate ; giving the highest curie temperature .
the eft and pa are much more accurate methods .
noticeably , the numerical calculations performed on the finite clusters with periodic boundary conditions seem to be the most accurate . as pointed out in the previous section , in this approach the curie temperatures have been identified from the maxima of the specific heat , according to eq.(13 ) . in fig.3 the specific heat curves per lattice site for finite clusters
are illustrated vs. temperature for various concentrations @xmath4 .
the temperatures corresponding to the maxima of these curves are denoted by the circular markers in fig.2 .
although the location of the specific heat peaks can be determined numerically from the curves , we found that the application of the analytical formula ( 13 ) leads to more precise results .
it should be noted that the specific heat in fig.3 behaves correctly from the thermodynamic point of view , both in the low and high temperature limits .
in the paper four approximate methods have been applied in order to study the low dimensional pt ising ferromagnet with selective dilution . within those methods the formulas for the curie temperatures have been obtained .
the phase diagram has been calculated for various values of the concentration parameter @xmath4 and the results of the different methods have been compared .
the most accurate method is the one based on the exact numerical calculations for finite clusters with the periodic boundary conditions .
it is worth noticing that within this method all thermodynamic properties can be simultaneously calculated in the same computational time .
however , regarding the determination of the curie temperature , this method is applicable to the systems in which the maximum of the specific heat is unambiguously related to the phase transition temperature .
this is not always the case ; for instance , for the frustrated systems or even some paramagnets where the so - called schottky maximum is observed . therefore , the analytical methods , giving better physical insight and proper interpretation of the numerical results , are still important in such studies . among analytical methods the pa approach can be especially recommended , for it enables the self - consistent studies of all thermodynamic properties based on the gibbs potential .
it also gives satisfactory accuracy when compared with other approaches .
as shown recently , this method can be applied to the heisenberg systems as well [ 24 , 25 ] .
it should also be noted that for quantum systems the numerical diagonalization of the finite cluster hamiltonian is much less efficient than ecfc for the classical ising model . for the model in question the antiferromagnetic interactions can also be considered .
then , for selective dilution @xmath57 the frustrations will occur and the theoretical description becomes more complex .
this problem should be a subject for separate paper .
a.m.luchli , in : _ introduction to frustrated magnetism : materials , experiments , theory _ , ed .
c.lacroix , p.mendels , f.mila , springer series in solid state sciences , vol .
164 ( springer , berlin , heidelberg 2011 ) , p. 481 | in the paper the curie temperatures of selectively diluted planar ising ferromagnet on the triangular lattice are calculated vs. concentration of magnetic atoms .
various analytical approaches are compared with the exact numerical calculations for finite clusters , as well as with the exact analytical solutions for the triangular and honeycomb lattices . |
a treatment of disorder is an essential part of the theory of superconductivity .
after all , one must explain why impurity scattering does not cause resistance .
thus it is natural that as evidence for novel superconducting states multiplies the foundations of the subject , due mainly to anderson @xcite and abrikosov and gorkov@xcite , are being re - examined .
the experiments which stimulate most strongly the current revival of interest in the problem are those on the high temperature superconductors@xcite , which are now universally regarded as ` @xmath1wave superconductors'@xcite , and those involving some of the heavy fermion systems which display signs of ` @xmath5wave ' pairing@xcite . in what follows
we wish to contribute to the theoretical discussion @xcite of the issues raised by these very interesting developments .
the case of classic , ` @xmath0wave ' , superconductors is by now well understood .
if the perturbation does not break time reversal symmetry and the coherence length is sufficiently long , so that the pairing potential @xmath6 does not fluctuate , the anderson theorem@xcite guarantees that there is an absolute gap in the quasi - particle spectrum and the main effect of disorder is that the density of normal states in the gap equation is replaced by its average over configurations@xcite . on the other hand
if the perturbation breaks time reversal invariance , as is the case with paramagnetic impurities , the effect is more dramatic .
for instance , the transition temperature @xmath7 is reduced from its clean limit value @xmath8 , according to the well known pair - breaking formula : @xmath9 where @xmath10 is the digamma function and @xmath11 is a measure of the strength of the scattering and @xmath12 is the scattering rate@xcite .
by contrast in the case of superconductors whose cooper pairs are of exotic ` @xmath5wave ' or ` @xmath1wave ' character even simple potential scattering , which does not break time reversal symmetry , causes pair breaking@xcite .
this fact was noted already in the early contributions to the field@xcite , but has become a subject of intense scrutiny only recently@xcite .
of particular interest are two dimensional models featuring ` @xmath1wave ' pairing as these may be relevant to experiments on high @xmath3 superconductors .
notably , for cuprates many experiments have explored the variation of @xmath3 , the density of states and other properties as a function of ni and zn substitutions on the copper sites@xcite or irradiation damage@xcite .
although a wide variety of theoretical ideas@xcite and methods@xcite have been applied to interpret the experiments a comprehensive picture of the role of disorder is far from complete . on a more formal level
, an intriguing problem arises from the observation of gorkov and kalugin@xcite that the scattering in models where the order parameter has a line of zeros on the fermi surface is highly singular and this may be a manifestation of interesting new physics . indeed in 2-d ,
et al._@xcite predict that the quasi particle density of states n(e ) approaches zero , even in the disordered state , as power low , @xmath13 , with positive exponent @xmath14 , instead of going to a finite value @xmath15 where @xmath16 measures strength of the interaction , as was found by gorkov and kalugin@xcite .
another interesting and controversial issue is the relative importance of the self - consistent born approximation ( scba ) and resonant scattering in the unitarity limit@xcite .
our aim here is to explore the subject systematically on the bases of explicit calculations , albeit for a simple , extended hubbard model with attractive interactions and site diagonal randomness only . in short
we will examine the problem of disordered unconventional superconductors making use of the coherent potential approximation ( cpa ) .
the cpa is the most reliable approximation developed for the theory of electronic structure of random metallic alloys in the normal state@xcite .
notably it has been shown to be exact in both the weak and the strong scattering limits , and applicable to systems with low as well as high concentration of impurities .
significantly , the cpa reduces to the self - consistent born approximation ( scba ) for weak scattering impurities , and agrees with the self - consistent t - matrix approximation ( scta ) results for strongly scattering impurities of low concentrations .
indeed it remains a good approximation in the unitarity limit of resonant scattering@xcite . finally , on account of the fact that it becomes exact as the number of nearest neighbors goes to infinity ,
the cpa is often referred to as a mean field theory of disorder@xcite .
given these desirable features it is clearly worthwhile to explore the consequences of the cpa for disordered superconductors .
for the case of conventional @xmath0wave pairing this has already been done , generating many useful results@xcite . apart from our earlier brief report@xcite and the limited discussion in ref . , the case of superconductors with cooper pairs of @xmath1 symmetry will be treated here , within cpa , for the first time .
we will describe fully our numerical method , demonstrate that in various limits our formalism reproduces many of the well known results for disordered superconductors , and examine in detail the phase diagram of the local and non - local attractive two - dimensional hubbard models .
in particular we study the variations of @xmath3 with impurity scattering strength and with impurity concentration for the case of local @xmath0wave pairing as well as non - local ( extended ) @xmath0wave and @xmath1wave pairing .
we also contrast the cases for a binary alloy , a - b type , disorder with the case of uniformly distributed scattering potentials on each site .
finally , we investigate the dos , @xmath17 , at low energies and its consequences for measurements of the specific heat , comparing our results with those of kalugin and gorkov@xcite , eq .
[ eqwenger ] , and of nereseyan , tsvelik and wenger @xcite .
our starting point is the single band hubbard model with an attractive extended interaction which is described by the following hamiltonian @xmath18 where @xmath19 and @xmath20 are , respectively , the usual creation and annihilation operators for electrons on site @xmath21 with spin @xmath22 , and the local charge operator is @xmath23 with @xmath24 .
the chemical potential is @xmath25 , @xmath26 are the hopping integrals ( for @xmath27 ) and @xmath28 is the local site energy . the interaction term , @xmath29 , can be either be a local attractive interaction ( @xmath30 ) giving rise to @xmath0wave pairing , or a non - local attractive interaction ( @xmath31 for @xmath27 ) giving rise to @xmath1wave or extended @xmath0wave pairing . disorder is introduced into the problem by allowing the site energies , @xmath28 , to vary randomly from site to site . starting from eq .
( [ eq : h ] ) we apply the hartree - fock - gorkov @xcite approximation , which results in the following bogoliubov de gennes equation : @xmath32 for the greens function matrix @xmath33 at the mastusbara frequency , @xmath34 . for computational convenience
we shall take the hopping integral @xmath26 to be non zero only when the sites @xmath21 and @xmath35 are nearest neighbors .
the mean field pairing potential @xmath36 can either be local ( @xmath37 ) or nearest neighbor non - local , see fig . 1 , and couples particle and hole amplitudes between sites @xmath21 and @xmath35 .
of course , the above equations are completed by the self - consistency condition that @xmath38 where @xmath39 is a positive infinitesimal . to simplify matters
we have assumed that the normal hartree and exchange terms can be absorbed into the definitions of the chemical potential , @xmath25 , or the hopping integrals @xmath26 . as usual , eqs .
( [ eq : bdge1 ] ) and ( [ eq : selfcon1 ] ) are to be solved subject to the requirement on the chemical potential that @xmath40 where @xmath41 is the average number of electrons per unit cell . clearly , the greens function matrix @xmath42 determined by the above equations depends on the set of site energies @xmath43 .
our task is to find the configurationally averaged greens function matrix @xmath44 .
evidently , this is made much easier if we assume that the pairing potential does not fluctuate from configuration to configuration . as was argued by gyorffy _
et al._@xcite this is a good approximation when the t=0 coherence length @xmath45 is large .
thus our results will have to be treated with appropriate care when applied to superconductors with short coherence length such as superconducting cuprates .
let us now proceed to deploy the cpa strategy for calculating the averaged greens function matrix @xmath44 subject to the self consistency conditions : @xmath46 @xmath47 the first move in deriving the fundamental equations of the coherent potential approximation is to define a coherent medium greens function matrix @xmath48 by @xmath49 as will be clear latter @xmath50 and hence @xmath51 and @xmath52 are the diagonal components of the usual self - energy .
note that we did not introduce any off diagonal self - energies such as @xmath53 and @xmath54 because for the single site perturbations of our model they are zero .
the next step is to consider the scattering of the quasi - particles , propagating according to @xmath48 by the defects described by the potentials : @xmath55 where @xmath56 labels one of the @xmath57 different site energies we wish to consider . in a straightforward application of the cpa principles,@xcite @xmath58 and therefore @xmath48
is determined by the condition that these defects do not scatter on the average _ i.e. _ @xmath59 where @xmath60^{-1}\ ] ] and @xmath61 from eqs .
( [ eq : zeroscat],[eq : scatmat ] ) it is now possible , in conjunction with equations ( [ eq : selfcon2]-[eq : bdge2 ] ) , to calculate @xmath58 and @xmath48 .
the numerical methodology for calculating @xmath48 and @xmath58 closely follows that in ref . and
is outlined in appendix a. a number of recent studies of superconductors with unconventional pairing suggest that the consequences of disorder depend sensitively on the models used to describe the randomness.@xcite thus we are going to investigate two different models .
the first corresponds to binary alloy disorder , where @xmath62 .
namely we consider two _ types _ of sites with site energies @xmath63 and @xmath64 and concentrations of @xmath65 and @xmath66 respectively .
the second model is described by a uniform distribution of site energies . here
we shall have in mind the limit where @xmath67 with @xmath68 $ ] .
consequently in eq .
( [ eq : zeroscat ] ) the sum @xmath69 becomes the integral @xmath70 . in the bimodal case , where @xmath62
, we can simplify eq .
( [ eq : zeroscat ] ) to find @xmath71 where @xmath72 . in the second , uniform distribution ,
case we transform the sum in equation ( [ eq : zeroscat ] ) into an integral so that @xmath73 after some straightforward algebra this leads to @xmath74 thus our cpa calculations will consist of solving numerically either eq .
( [ eq : sig11_1 ] ) for the bimodal distribution of the site energies , or eq .
( [ eq : sig11_2 ] ) for the case of uniform distribution , to determine the self - energies @xmath51 and @xmath75 .
we now relate the cpa formulae derived above to the usual results of disordered superconductors , corresponding to the well known pair breaking formula eq .
( [ eqkalugin ] ) . as is well known@xcite the pair breaking formula
was first derived for magnetic impurities in @xmath0wave superconductors@xcite but it also applies in many other interesting circumstances such as our present concern , namely the case of non - magnetic impurities in @xmath1wave superconductors@xcite . to derive it within the cpa
let us start with the gap equation @xmath76 as a motivation for our argument we recall the method of abrikosov and gorkov@xcite for solving the gap equation at @xmath3 for a clean superconductor . in that case , to find @xmath3 we linearize the analogue of eq.([eq : gap1 ] ) by approximating the off diagonal greens function @xmath77 as follows : @xmath78 where @xmath79 , and for our tight binding model with a square lattice @xmath80 .
then , we note that the kernel of the linear integral equation for @xmath81 is a four term degenerate kernel : @xmath82 where @xmath83 and @xmath84 .
consequently , the general @xmath85 will be a linear superposition of @xmath86 and @xmath87 .
however , when the internal symmetry of the singlet cooper pair is pure @xmath1wave we may take @xmath85 to be of the form @xmath88 then the condition for non - zero order parameter becomes @xmath89 let us now define a @xmath1wave weighted density of states @xmath90 and write the above condition , which determines the transition temperature @xmath8 , as @xmath91 where @xmath92 .
in the above equation the integral and the sum are divergent , so we need to introduce a cut - off , @xmath93 . in the usual way we assume the density of states @xmath94 is slowly varying up to the cut off energy ,
so we will make the approximation @xmath95 .
then , considering that @xmath96 we can write @xmath97 and hence rewrite eq .
( [ eq : transtemp ] ) as @xmath98 this is the bcs result for the superconducting transition temperature in the case of @xmath1wave pairing@xcite .
it differs from the conventional result only in that the @xmath1projected density of states @xmath99 has replaced the usual full density of states @xmath100 .
let us now return to disordered superconductors and examine how the above well known argument is modified when the randomness is dealt with within the cpa . using eq .
( [ eq : bdge2 ] ) it can be easily seen that instead of eq .
( [ eq : gaptc ] ) we should use @xmath101 to linearize eq . ( [ eq : gap1 ] ) at @xmath3 .
thus , noting that @xmath102 the condition which determines @xmath3 can be written as @xmath103 now , at this point we need to know the form of @xmath51 to progress any further .
as a first approximation we assume that the most important component to the self - energy is the component at the fermi energy @xmath104 . later on we will test the accuracy of this approximation by examining our numerical results for @xmath51 . for now , however , let us proceed by taking @xmath105 evidently this leads to @xmath106 again taking @xmath94 outside of the integration as @xmath99 and performing the integration over @xmath107 we find @xmath108 where again the sum is cut off , as in the clean limit , by @xmath93 . if we now add and subtract the terms corresponding to @xmath109 ( the clean case ) we find @xmath110 clearly the term @xmath111 on the lhs of eq .
( [ eq : addsub ] ) can be replaced by @xmath112 on account of eq .
( [ eq : clean ] ) . with the same accuracy
the first sum on the rhs of equation ( [ eq : addsub ] ) equals @xmath113 and the second sum is convergent .
hence the cutoff @xmath93 can be extended to infinity . as has been noted frequently before this infinite sum can be readily performed@xcite and we find @xmath114 where @xmath115 eqs .
( [ eq : pairbreak ] ) and ( [ eq : rhoc ] ) are the central results of this section .
reassuringly , whilst equation ( [ eq : pairbreak ] ) is the standard pair - breaking formula,@xcite eq . ( [ eq : rhoc ] ) is a very natural , but novel , cpa expression for the pair breaking parameter @xmath116 . recall that our derivation of the above result from cpa involved the approximation : @xmath117 . to test the validity of this approximation we wish to compare exact cpa numerical results with the predictions of the analytical expression : eqs .
( [ eq : pairbreak ] ) and ( [ eq : rhoc ] ) .
using numerical solutions of the cpa equation , to be discussed latter , fig .
2 plots the pair breaking strength @xmath116 vs. @xmath118 , the disorder strength for the binary alloy type disorder . to find pair breaking parameter @xmath116
we calculate @xmath3 for each disorder strength , @xmath118 , and inverted eq .
( [ eq : pairbreak ] ) to obtain an effective @xmath116 .
the exact cpa @xmath116 can then be compared to the solid line in fig . 2 where we have taken our numerically calculated values for @xmath119 and directly calculated @xmath116 , via eq .
( [ eq : rhoc ] ) .
finally the dashed line in fig .
2 corresponds to @xmath116 obtained using the self - consistent born approximation ( scba ) .
evidently , the self energy at the fermi energy , @xmath120 , @xmath119 , gives a good description of the pair breaking parameter @xmath116 via eq .
( [ eq : pairbreak ] ) .
also it is clear that , as expected , the self - consistent born approximation @xmath121 only works well in the weak scattering limit .
in this section we examine various analytically accessible limits of the cpa formalism described above .
firstly , we demonstrate that anderson s theorem is obeyed for @xmath0wave superconductors and the cpa equations are consistent with the results of abrikosov and gorkov@xcite .
secondly we show that for @xmath1wave superconductors the quasi particle density of states at the fermi energy , @xmath100 , is non - zero in the presence of non magnetic disorder scattering and is consistent with the gorkov - kalugin formula , eq .
( [ eqwenger ] ) . formally , the cpa eqs .
( [ eq : zeroscat ] ) , ( [ eq : scatmat ] ) can be written in terms of renormalized matsubara frequencies , @xmath122 , pairing parameter , @xmath123 , and particle energies @xmath124 .
this quantities are defined as follows @xmath125 consequently @xmath126 and for the alloy type disorder with @xmath127 and @xmath128 , the self energy @xmath129 which renormalizes @xmath130 and @xmath131 , and is defined by eq .
( [ eq : sig11_1 ] ) can be written as @xmath132 the alternative expression for @xmath51 in the case of a uniform distribution of local potentials , @xmath133 is given in eq .
( [ eq : sig11_2 ] ) .
note that in the case of a non isotropic , @xmath1wave gap , @xmath134 , is not renormalized if the disorder is diagonal both in site and nambu space .
this is different from the case of local @xmath0wave pairing where , as can be readily shown , @xmath134 is renormalized by the same factor as @xmath122 in eq .
( [ eq : renorm ] ) .
thus for conventional superconductors , in contradiction to eqs .
( [ eq : renorm ] ) - ( [ eq : g_11 ] ) , we find that the cpa yields @xmath135 in agreement with born approximation or abrikosov - gorkov theory@xcite . as is widely appreciated @xcite the above equation implies the anderson s theorem in @xmath0wave superconductors .
by contrast in the @xmath1wave case represented by eqs .
( [ eq : renorm ] ) , eq.([eq : s_renorm ] ) does not hold and hence there is no anderson theorem .
finally in concluding this section we would like to stress that eqs .
( [ eq : renorm ] ) - ( [ eq : sig113 ] ) represent strictly pure @xmath1wave result .
even if we stick to the singlet case a more general solution of the cpa equation will imply a renormalization of @xmath134 to @xmath136 .
a good example of such situation is a case where the symmetry of @xmath137 is of extended @xmath0wave symmetry @xmath138 type
. we shall encounter this interesting circumstance later on in this paper . moving on and returning to the @xmath1wave case , we observe that the form of eqs .
( [ eq : renorm ] ) - ( [ eq : sig113 ] ) are the same as were found by larkin.@xcite thus again the cpa reproduces the expected general form of the gap and frequency renormalizations , but with an improved description of the disorder .
the most prominent feature of a conventional superconductor is vanishing of the quasi particle density of states @xmath139 for energies @xmath140 measured from the fermi energy @xmath141 , less then @xmath6 . in the case of clean , @xmath1wave superconductors the line of zeros of @xmath85 on the fermi surface leads to finite @xmath139 for all @xmath140 except @xmath142 .
in fact as is well known@xcite @xmath139 approaches zero linearly in @xmath140 . in the present section
we shall investigate what happens to @xmath139 in the presence of disorder .
as it turns out for a given gap parameter @xmath143 and in the limit of small disorder @xmath144 the cpa equations can be solved analytically . to affect the solution note that in eq .
( [ eq : g_11 ] ) the major part of the summation is coming from the four singular points in the brillouin zone where the denominator vanishes : @xmath145 linearizing around these points and performing the summation over @xmath146 analytically , we find that @xmath147 where @xmath148 .
clearly in the limit @xmath149 this leads to @xmath150 moreover , when @xmath51 is small compared to the band width we can rewrite eq .
( [ eq : sig113 ] ) , using the self - consistent born approximation as @xmath151 on substituting this result into eq .
( [ eq : g0 ] ) the later becomes an equation for @xmath152 which determines the density of quasi particle state @xmath100 _ via _ the formula @xmath153 .
indeed we find @xmath154 and @xmath155 these formulae agree with the results of kalugin and gorkov@xcite and haas _
_ @xcite and have been verified numerically .
for example figs . 3 and 4 compare @xmath100 as calculated from eq .
( [ eq : dos ] ) and calculated using completely self - consistent cpa .
as one can see there is good agreement between the two results and thus we conclude that at @xmath156 the density of states , in cpa , becomes finite when an arbitrary small amount of disorder is introduced .
however , one should note that at very low levels of disorder our numerical results tend to a constant , whereas the analytical results tend to zero , this comes from the small complex component we have added to the energy to enable us to evaluate the greens function . for a more detailed description of what is happening to @xmath100 and @xmath157 , in the presence of disorder , see sec .
vii of this paper , where we have analyzed their properties more closely and report , extensively , further numerical results .
in this section we present results for the single site local quasi particle density of states calculations .
as mentioned before we consider two types of disorder , ( i ) binary alloys where we have two on - site potentials randomly distributed throughout the lattice and ( ii ) a uniform distribution of random on - site potentials . for both types of disorder
we have solved , numerically , eq .
( [ eq : bdge2 ] ) in conjunction with its appropriate self - consistency conditions , for the order parameter , @xmath6 , the average number density , @xmath41 , and the self - energy , @xmath158 .
the first situation we consider is a binary alloy of random on site energies @xmath63 and @xmath64 with equal concentration @xmath159 .
the parameter we have chosen to use to describe the strength of the disorder is @xmath160 .
figure 5(a ) shows the density of states in the normal state .
the van hove singularity characteristic of a tight - binding model with nearest neighbor hopping on a square lattice is clearly visible for small disorder ( @xmath161 ) in the middle of the band . as fig .
5(a ) shows , for more strongly disordered alloys this van hove peak is split into two peaks with some additional smearing . in the limit of @xmath162 being very large
we get , as one would expect , band splitting of states associated with @xmath63 and @xmath64 respectively . on the other hand the disorder with uniform distribution of site energies , fig .
6(a ) , gives only the smearing and flattening of density of states with , eventually , a complete flattening of the van hove peak . the imaginary part of the self - energies for these two types of disorder are shown in figs .
5(b ) and 6(b ) respectively . evidently at low levels of disorder
, they reflect the density of states of the clean system , and hence are consistent with the born approximation _ via .
_ @xmath163 . for larger disorder strengths @xmath164 is more flat for the uniform distribution , whereas for the binary alloy the self - energy is peaked not at the van hove singularity but at @xmath63 and @xmath64 . before turning to the problem of disordered @xmath1wave superconductors for reference we have studied , briefly , the @xmath0wave case . in short ,
we have introduced a site diagonal , local , attractive interaction with strength @xmath165 , into the above model .
as expected such interaction leads to conventional @xmath0wave pairing . when implementing the cpa we have assumed that in this case the pairing order parameter @xmath6 , which is now site diagonal does not vary from site to site .
this is consistent with the anderson theorem which was shown to be adequate for systems with a long coherence length.@xcite fig .
7(a ) shows the @xmath0wave quasi - particle density of states as calculated for various values of the binary alloy disorder strength @xmath162 .
the results confirm that , for @xmath0wave pairing , the gap is absolute and , whilst the edges may move , they remain well defined as required by anderson s theorem @xcite . in fig .
7(b ) we have also plotted the self - consistent self - energy as a function of the quasi - particle energy , for the same disorder strengths used to obtain fig 7(a ) . from this graph
we can see that @xmath166 is zero inside the gap and hence the normal disorder does not act as a pair breaker , for @xmath0wave superconductivity .
the quasi - particle density of states for @xmath1wave superconductors is dramatically different from the above bcs spectrum in the @xmath0wave case even in the clean limit .
as it is well known it has the characteristic v - like dip shown in fig .
8(a ) , for @xmath140 near the chemical potential ( @xmath167 ) . upon introducing binary alloy disorder into the @xmath1wave system one would naturally expect similarly different behavior from that described above and in fig .
7 . as is clear from the results reported in fig .
8 this is indeed the case .
strikingly , as the analytic results of the previous section suggested , n(0 ) becomes non - zero for the slightest disorder .
this implies gapless superconductivity in contrast to the gaped one in the @xmath0wave case . in fig .
8(b ) we have plotted the self - energies , corresponding to the quasi - particle density of states results presented in fig .
this shows a complex evolution with disorder , for small @xmath162 the imaginary part of the self - energy , @xmath168 , reflects the pure @xmath1wave density of states ( as expected in the scba limit ) and hence @xmath169\sim 0 $ ] .
however increasing @xmath162 leads to a finite @xmath170 , with a cusp like minimum in @xmath171 $ ] at @xmath172 . increasing the disorder even further , to @xmath173 , the @xmath1wave pairing
is completely destroyed , and @xmath174 reverts to the normal system self - energy . in this case
@xmath175 is a maximum at @xmath172 , since the fermi energy , @xmath176 , was set exactly to @xmath177 .
to analyze the effect of the disorder upon @xmath3 we have solved the gap equation together with the cpa equations .
again for the sake of comparison we have developed the analogous theory for the conventional @xmath0wave superconductors based on the site diagonal particle - particle interaction , of strength @xmath178 , mentioned briefly in a previous sections . in this case neglecting the spatial fluctuations of @xmath179 , and linearizing the gap equation for the configurationally averaged single site pairing potential @xmath180 we find the condition for the temperature @xmath3 , below which @xmath181 , to be @xmath182 where @xmath183 is the averaged density of states in the normal state at energy @xmath107 , as calculated within the cpa , and @xmath184 .
the solution of eq .
( [ eq : tc52 ] ) for @xmath3 as a function of band filling and strengths of alloyed disorder , @xmath162 is shown in fig .
9 . from this figure
we can see that the superconducting state exists for all band filling and the strength of disorder does little to suppress @xmath3 .
again this is consistent with anderson s theorem .
now let us turn to the case where the interaction is non - local and the pairing potential @xmath36 connects nearest neighbors sites .
near @xmath3 , where the gap equation is linear , the solutions can be labeled by their symmetries . indeed we find two separate conditions on the temperature @xmath3 for the instability of the normal state to @xmath1 and extended @xmath0 symmetry breaking . in the first case @xmath185
is determined by @xmath186 and in the second extended @xmath0wave case , @xmath187 is given by @xmath188 where @xmath189 and @xmath190 refer to @xmath0wave and @xmath1wave like ` harmonics ' defined previously following eq .
[ eq : pot ] . in fig .
10 the critical temperature for both @xmath1wave and extended @xmath0wave pairing is shown as a function of band filling , @xmath41 , for various strengths of alloyed disorder , @xmath162 .
full lines correspond to @xmath1wave @xmath3 while dashed ones to extended @xmath0wave solutions . in the clean limit ( @xmath191 )
we can see that the extended @xmath0wave solution exists mainly at the band edges and the @xmath1wave solution is largely confined to the central portion of the band . for the interaction strength we have chosen , the two solutions cross at @xmath192 .
evidently for @xmath193 the superconducting instability is at @xmath187 whilst @xmath194 the transition temperature is @xmath185 as we increase the strength of the alloyed disorder ( @xmath195 from top to bottom curve ) we see that both the @xmath3 for the extended @xmath0wave and @xmath1wave solutions is reduced and for particularly strong disorder ( @xmath196 and @xmath197 ) the maximum in the @xmath1wave @xmath3 is no longer at @xmath198 .
this is connected with the splitting of van hove singularities visible in fig .
5(a ) , _ i.e. _ the maximum in the normal - state density of states corresponds to the maximum in the @xmath1wave @xmath185 . on the other hand for a uniform distribution of random site energies where there is no splitting of van hove singularities , and
hence , we see a different tendency ( fig . 11 , @xmath199 from top to bottom curve ) .
the maximum value of @xmath185 is located at @xmath198 no matter what the strength of disorder . for extended @xmath0wave solutions and for relatively small disorder in the of @xmath1wave case the decreasing tendency in @xmath200 with growing disorder
are very similar for the two types of disorder . however above a certain strength in either case
we note that there is no crossing between the extended @xmath0wave and @xmath1wave solutions .
this fact may be interpreted as the sign that the @xmath0 and @xmath1 sates can not coexist or mix . to compare the effect
the two types of disorder have upon @xmath3 we have , in fig .
12 , calculated @xmath185 vs. @xmath118 at half filling for both binary alloy ( solid line ) and uniform ( dashed line ) disorder .
from this graph we can see that the two types of disorder effect @xmath3 in almost exactly the same manner .
if we rescale the graphs such that the standard deviation in the disorder strength is the same , for uniform disorder @xmath201 and for binary disorder @xmath202 , then the two curves lie very close to one another .
this prompts the conclusion , which may however be premature,@xcite that @xmath3 is not very sensitive to the details of the fluctuation in the site energy . other interesting point to investigate concerns the relative robustness of @xmath3 against degradation by disorder in the cases of conventional @xmath0 and @xmath1wave pairing.@xcite fig .
13 shows the results for @xmath3 for inter - site @xmath1wave and on site @xmath0wave superconductors versus alloyed disorder strength @xmath162 , where @xmath203 . here
one can easily recognize a typical difference between these two superconducting states . clearly in case of a @xmath1wave superconducting state disorder acts as an effective pair breaker while for @xmath0wave superconductors it decreases @xmath3 only slowly or not at all .
an interesting physical consequence of this effect is that if both @xmath204 and @xmath205 ( @xmath206 ) are attractive increasing disorder could lead to a transition from @xmath1wave to @xmath0wave pairing , as suggested by abrikosov@xcite .
evidently , the binary alloy disorder has certain features not shared by the uniform distribution of site energies model .
one of these is the possibility of band splitting . for a very large @xmath162 parameter the band
is split into two sub - bands , as shown in fig .
the on site @xmath0wave superconducting state is created within one of sub - bands . for disorder parameters : @xmath127 and @xmath207 we have not found any similar solution for @xmath1wave , but for c=0.25 and n=0.5 a single sub - band @xmath1wave solution is found . in fig .
15 we show the very different behavior of @xmath3 for a small and a large non - local interaction parameter , @xmath208 .
for large enough @xmath208 the @xmath1wave @xmath3 does not go to zero when the band splits .
the quasi - particle density of states corresponding to such a possibility is presented in fig .
16 and , as in fig . 14 , we see the superconducting gap , which is now @xmath1-wave , has formed in one of the sub - bands . to complete the discussion figs .
17 , 18 and 19 show the critical temperature plotted versus concentration c for various strength of binary alloy disorder @xmath162 with band fillings of @xmath198 ( fig .
17 ) , @xmath209 ( fig . 18 ) and @xmath210 ( fig .
19 ) . figs .
17 and 18 correspond to @xmath1wave while fig .
19 correspond to inter - site extended @xmath0wave pairing .
again for large enough alloy disorder pair breaking phenomena take place .
if disorder is introduced by a fixed potential @xmath162 , but varying the concentration of scatters , @xmath211 , from @xmath177 we see that the results are similar to the case of a fixed @xmath65 and increasing @xmath162 from @xmath177 .
this shows that cpa can work equally well , and indeed asymptotically exact , in both the born ( small @xmath162 ) and @xmath2matrix ( small @xmath65 ) and resonant scattering regimes .
as mentioned earlier the behavior of the quasiparticle density of states @xmath17 and the imaginary part of the self energy @xmath214 near @xmath215 is of general conceptual significance .
for instance , the power law behavior of @xmath216 , instead of the exponential cut off characteristic of a gap in the quasiparticle spectrum.@xcite naturally , dramatic changes in the low energy behavior of @xmath139 and @xmath217 as disorder is added to the problem is also of general interest and , as it turns out , of lively controversy.@xcite in this section we wish to examine those predictions of the cpa calculations which are relevant to these issues in particular . using the methods outlined in the previous sections we calculate numerically the quasi - particle local density of states , @xmath139 , and the self - energy , @xmath218 , and
then investigate how these two quantities change with both temperature and disorder . for simplicity
we have studied the half filled band ( @xmath167 , @xmath198 ) .
this is the band filling for which @xmath3 , in the @xmath1wave case , is a maximum .
we will examine the effects of alloyed disorder on the system and show that as we increase the amount of scattering the specific heat vs. temperature relation changes from @xmath219 , in the clean limit , to @xmath2 in the limit of strong scattering .
as is widely recognized @xcite this behavior is the consequence of @xmath220 changing to @xmath221 and is consistent with experiments .
@xcite to get an impression of the form of @xmath17 and @xmath222 in the region of the chemical potential , @xmath25 , we fit them to the function @xmath223 in the energy range @xmath224 , where @xmath225 is small compared to the gap . using the coefficients @xmath226 , @xmath227 and @xmath65 to fit the curves @xmath17 and @xmath228 in the region of @xmath25
gives us a tool to analyze their functional form .
for example @xmath226 tells us if the curves are finite at @xmath172 , @xmath227 controls the gradient of the curve and @xmath65 controls the curvature . in figs .
20(a - c ) we have plotted these coefficients , for @xmath17 , as a function of @xmath2 , for different values of @xmath162 .
we also included in fig .
20(d ) plots of the magnitude of the @xmath1wave superconducting order parameter @xmath229 vs. @xmath2 , for the same values of alloyed disorder strength , @xmath162 , to indicate the temperature where the order parameter goes to zero . regarding these curves as a brief summary of what the cpa predicts about the low energy behavior of @xmath17 and @xmath168 we now comment on their implications .
firstly we note that in fig .
20(a ) the parameter @xmath230 tends to a finite limit as @xmath231 and this limit increases more and more rapidly as the disorder described by @xmath162 , increases .
hence @xmath100 is finite in agreement with our earlier discussion in sec .
iv where we derived for low scattering and low temperatures the dependence of @xmath100 on @xmath162 ( see eq .
( [ eq : dos ] ) ) .
it lends credit to the general consistency of our results that the @xmath232 curves rise to their normal state values as @xmath233 .
interestingly , as can be seen in fig .
20(b ) the gradient of @xmath17 , namely @xmath227 , changes dramatically only near @xmath3 .
in fact for the case where @xmath162 is small ( squares ) the gradient even changes sign .
this means that we go from having a gap in the density of states , below @xmath3 to having a van hove singularity above @xmath3 .
the curvature , represented by @xmath65 and plotted in fig .
20(c ) , shows that for low disorder ( squares ) as we increase the temperature we go from a curvature of @xmath234 , ie .
a cusp , to a curvature @xmath235 .
finally when the critical temperature is reached for each of the four different disorder types @xmath65 goes to @xmath236 . in the same manner
, we now examine the corresponding coefficients , @xmath237 , @xmath238 and @xmath239 for @xmath228 . in figs .
21(a - c ) we have plotted the calculated coefficients for @xmath228 , at different values of @xmath162 as a function of the temperature . in fig .
21(a ) we can see that at low temperatures @xmath157 increases with disorder , again this agrees with our results in sec .
iv where an analytical form for the dependence of @xmath240 upon @xmath162 was derived at @xmath241 , see eq .
( [ eq : sig0 ] ) .
as we increase the temperature we can see the self - energy at the chemical potential also rises until the normal state value is reached at the critical temperature . in fig .
21(b ) we can see how the gradient changes from being large for large @xmath162 and small for small @xmath162 at low @xmath2 , and at temperatures greater than @xmath3 to be large and negative for large @xmath162 and small and negative for small @xmath162 . finally fig .
21(c ) shows how , @xmath65 , the curvature goes from being almost linear , @xmath242 , at low temperatures , regardless of @xmath162 , to @xmath243 above @xmath3 . to illustrate the consequences of these results for the low temperature thermodynamic properties
we have calculated the effect of disorder upon the specific heat .
using the above temperature dependent coefficients for the density of states near @xmath244 we calculated the limiting behavior of the specific heat as @xmath245 .
the results are shown in fig .
22 for different strengths of disorder @xmath162 . in the case where @xmath162 is small ( squares ) we can see that the specific heat has a @xmath219 dependence ( the plotted continuous solid line with no points ) and for large @xmath162 the dependence upon temperature is linear , as expected.@xcite
we have compared and contrasted the effect of disorder on conventional @xmath0 and @xmath1wave superconductors on the bases of an extended , negative @xmath165 hubbard model and a mean field , cpa , treatment of disorder . on the one hand
we have derived many of the well known results , such as the pair breaking formula in eq .
( [ eq : pairbreak ] ) or that for the quasi particle density of states @xmath100 , eq .
( [ eq : dos ] ) . on the other hand
we have solved the gorkov
cpa equations eqs .
( [ eq : sig11_1 ] ) , ( [ eq : gap1 ] ) numerically and surveyed the salient features of their consequences by explicit calculations .
the use of cpa in describing disordered @xmath1wave superconductors is an advance in this very active field @xcite because it allows us to avoid the usually delicate choice between methods , sets of diagrams , designed to deal with either weakly or strongly scattering perturbations .
the cpa treats both kinds of problems equally accurately and it is known to provide a very credible interpolation between the two@xcite .
as an example where above feature of cpa may have a crucial role to play we recall the use of the resonant scatterer model in interpreting experimental data both on the cuprates @xcite and some heavy fermion systems.@xcite in short , note that in the impurity , @xmath246 , limit the self energy for the greens function describing an electron moving on a lattice , is given by @xmath247 where the @xmath2matrix is defined , in terms of the impurity ` potential ' @xmath248 , a real local shift in the site energy , as @xmath249 evidently @xmath250 is complex number with an amplitude : @xmath251 and a phase , the phase shift @xmath252 , given by @xmath253 now , observe that while for weak scatterers @xmath254 for a resonant scatterer in the unitary limit , defined by @xmath255 , @xmath256 where @xmath257 .
it is the above dramatic difference in the dependence of @xmath258 and @xmath259 on the density of states @xmath17 that the cited authors rely on in interpreting the relevant experimental data .
evidently , since the individual scattering events described by the local @xmath2matrices are always treated exactly in the cpa , the cpa describes weak scatterers and resonant scatterers equally well .
moreover , since it is a reliable approximation not only in the impurity limit , @xmath260 , but also for arbitrary concentrations it deals with resonance scatterers even when eq .
( [ eq : sigma_r ] ) no longer holds .
thus cpa should be the preferred treatment for models with strong , even resonant scattering .
although we have not specifically concentrated on this aspect of the method , the principle feature of eqs .
( [ eq : sigma_b ] ) and ( [ eq : sigma_r ] ) , namely the dependence of @xmath261 on @xmath262 , can be seen to be at work in our calculations of the previous section . to demonstrate this we have calculated @xmath170 and n(0 ) as functions of band filling , @xmath41 , in the most interesting region @xmath263 , and compared in fig .
23 their relationships for weak and strong scatterers at @xmath127 . clearly , for weak scatterers @xmath264 , @xmath265 is a more or less linear function of @xmath100 as in eq .
( [ eq : sigma_b ] ) , while for strong scatterers @xmath266 , @xmath267 is inversely proportional to @xmath100 as in the resonant scatterers model described by eq .
( [ eq : sigma_r ] ) .
thus we conclude that the cpa employed in the calculation gives a reliable account of disorder in both the weak and strong scattering regimes .
having listed the above desirable properties of the cpa we hasten to emphasize that it is a ` mean field ' theory of disorder and , hence , does not describe such interesting phenomena as localization @xcite even in the normal state .
@xcite consequently , our result that @xmath100 is finite for the smallest amount of disorder can not be taken as evidence against the conclusion that @xmath268 of nersesyan _
et al._. @xcite as this originates from the divergence of the vertex corrections in perturbation series for @xmath168 we may conjecture that it has to do with localization effects not described by cpa .
this very interesting point is in the need of further clarification , and indeed we shall return to it in a later publication . this work has been partially supported by epsrc under grant number gr / l22454 and kbn grants 2p 03b 031 11 and 2p 03b 050 15 .
a.m. martin would also like to thank the tmr network dynamics of nanostructures .
the recursion method@xcite was first used for solving the bogoliubov de gennes equation by annett and goldenfeld@xcite and litak _
et al_@xcite . its use to implement cpa calculations was advocated by julien and mayo@xcite . here , as in reference,@xcite we made a combined use of these powerful methods to calculate the greens function matrix @xmath269 where the indices @xmath21 and @xmath35 denote sites , while @xmath14 and @xmath270 represent the particle or hole degree of freedom on each site .
we denote particle degrees of freedom by @xmath271 and hole degrees of freedom by @xmath272 . for example
@xmath273 represents the greens function between the particle degree of freedom on site @xmath21 and the hole degree of freedom on site @xmath35 at energy @xmath107 . to compute the greens functions ( [ eq:6 ] )
we can closely follow the method described by martin and annett.@xcite using this method we can transform the hamiltonian to a block tridiagonal form @xmath274 where @xmath275 and @xmath276 are @xmath277 matrices . given this form for @xmath278 and expressing the greens function as @xmath279 the greens functions above
can be evaluated as a matrix continued fraction so that @xmath280 where @xmath281 within eqs .
( [ eq:7 ] ) and ( [ eq:9 ] ) we have a formally exact representation of the greens functions .
however in general both the tridiagonal representation of the hamiltonian , and the matrix continued fraction ( [ eq:7 ] ) will be infinite .
in practice one can only calculate a finite number of terms in the continued fraction exactly . in the terminology of the recursion method it is necessary to _ terminate _ the continued fraction .
@xcite if we were to calculate up to and including @xmath275 and @xmath282 and then simply set subsequent coefficients to zero then the greens function would have @xmath283 poles along the real axis .
the density of states would then correspond to a set of @xmath283 delta functions . in order to obtain accurate results it would be necessary to calculate a large number of exact levels , which would be expensive in terms of both computer time and memory . as a more efficient alternative we choose to terminate the continued fraction using the extrapolation method , as used previously by litak , miller and gyrffy.@xcite we calculate the values for @xmath275 and @xmath276 exactly up to the first @xmath57 coefficients using the recursion method .
then , noting the fact that the elements of the matrices @xmath275 and @xmath276 vary in a predictable manner,@xcite we extrapolate the elements of the matrices for a further @xmath146 iterations , where @xmath146 is usually very much greater than @xmath57 .
this enables us to compute the various greens functions accurately with relatively little computer time and memory .
we now have a method for calculating the greens function @xmath284 .
to calculate our system of equations self - consistently for a given site @xmath21 , we need to know @xmath285 , @xmath286 , @xmath287 and @xmath288 .
if the interaction is non - local , else if the interaction is local then we only need to calculate @xmath289 , see fig . 1 . to calculate @xmath290
we also need @xmath291 . in performing our self - consistency calculation
we can make use of the fact that @xmath292 just acts as a renormalization to @xmath293 ( see sec .
this means we do not need to recalculate @xmath275 and @xmath276 when we calculate the self - consistent self - energy .
we first calculate the order parameter self - consistently for a given attractive interaction , then we calculate @xmath294 and hence the self - energy self - consistently ( remembering that in this self - consistent loop we do not need to recalculate @xmath275 and @xmath276 , since the self - energy does not affect them ) .
then we recalculate the superconducting order parameter self - consistently , incorporating the calculated self energy and repeat this procedure until both the order parameter and the self energy have reached stable self - consistent solutions .
hardy , d.a .
bonn , d.c .
morgan , r.x .
laing and k. zhang , phys .
lett . * 70 * , 3999 ( 1993 ) ; d.a .
bonn , s. kammal , k. zhang , r.x .
liang , d.j .
baar , e. kleinand w.n .
hardy , phys .
b * 50 * , 4051 ( 1994 ) .
nersesyan , a.m. tsvelik and f. wenger , phys .
lett . * 72 * , 2628 ( 1994 ) , a.a .
nersesyan , a.m. tsvelik and f. wenger , nuc .
b * 438 * , 561 ( 1995 ) , a.a .
nersesyan and a.m. tsvelik , phys .
* 78 * , 3981 ( 1997 ) .
s.haas , a.v .
balatsky , m. sigrist and t.m .
rice , phys .
b * 56 * , 5108 ( 1997 )
. a.p . kampf and t.p .
devereaux , phys .
rev . b*56 * , 2360 ( 1997 ) .
g. hara and a.d .
nagi , phys .
b*58 * , 12 441 ( 1998 ) .
b.c . den hertog and m.p .
das , phys .
rev . b*58 * , 2838 ( 1998 ) .
radtke , k.levin , h .- b .
schttler and m.r .
norman , phys .
b * 48 * , 653 ( 1993 ) . l.a .
openov , pisma zh .
fiz . * 66 * , 627 ( 1997 ) [ jetp lett . * 66 * , 661 ( 1997 ) ] .
byers , m.e .
flatte and d.j .
scalpino , phys .
lett . * 71 * , 3363 ( 1993 ) ; m.e .
flatte and j.m .
byers , _ ibid . _ * 78 * , 3761 ( 1997 ) .
r. haydock , in : solid state physics , vol .
* 35 * , eds .
ehrenreich , f. seitz and d. turnbull ( academic press , new york , 1980 ) .
annett and n. goldenfeld , j. low .
89 * , 197 ( 1992 ) .
g. litak , p. miller and b.l .
gyrffy , physica c * 251 * , 263 ( 1995 ) .
julien and d. mayo , j. phys f * 3 * , 1861 ( 1993 ) .
a. magnus in : solid state sciences * 58 * eds .
pettifor and d.l .
weaire ( spriger 1984 ) .
f. ducastelle , p. turchi and g. treglia in : solid state sciences * 58 * eds .
d.g . pettifor and d.l .
nex in : solid state sciences * 58 * eds .
d.g . pettifor and d.l .
weaire ( springer 1984 ) .
g. allan in : solid state sciences * 58 * eds .
d.g . pettifor and d.l .
weaire ( springer 1984 ) .
a. trias , m. kiwi and m. weissmann , phys .
b * 28 * , 1859 ( 1983 ) . | a coherent potential approximation is developed for @xmath0wave and @xmath1wave superconductivity in disordered systems .
we show that the cpa formalism reproduces the standard pair - breaking formula , the self - consistent born approximation and the self - consistent @xmath2-matrix approximation in the appropriate limits .
we implement the theory and compute @xmath3 for @xmath0wave and @xmath1wave pairing using an attractive nearest neighbor hubbard model featuring both binary alloy disorder and a uniform distribution of scattering site potentials .
we determine the density of states and examine its consequences for low temperature heat capacity .
we find that our results are in qualitative agreement with measurements on zn doped ybco superconductors .
# 1@xmath4#1 |
as is well known , in the case of point particles , short distance physics directly translates into high energy physics .
this is a simple consequence of the heisenberg uncertainty principle . in local quantum field theories , which describe the dynamics of point particles ,
the fundamental degrees of freedom are revealed at high energy , or equivalently , at short distance .
also , there is a clear separation between ultraviolet and infrared physics from the point of view of the renormalization group . in string theory , however , there is growing evidence that the physics at short distances , in contrast to local quantum field theory , is not clearly separated from the physics at long distances @xcite . the fundamental formulation of this
so called uv / ir mixing , as well as its observable consequences , are not understood at present .
various authors have argued that some kind of uv / ir mixing is necessary to understand the cosmological constant problem @xcite or the observable implications of short distance physics on inflationary cosmology @xcite .
motivated by these questions , we have recently @xcite investigated various observable consequences of the uv / ir mixing embodied in the `` deformed '' commutation relation @xcite @xmath1 = i\hbar ( 1 + \beta \hat{p}^2 ) \;. \label{eq : com1}\ ] ] this commutation relation implements the minimal length uncertainty relation @xmath2 which appears in perturbative string theory @xcite .
note the uv / ir mixing manifest in eq .
( [ uncert ] ) : when the uncertainty in momentum @xmath3 is large ( uv ) , the uncertainty in the position @xmath4 is proportional to @xmath3 and is therefore also large ( ir ) .
note also that eq .
( [ uncert ] ) implies a lower bound for @xmath4 : @xmath5 in the context of perturbative string theory , the existence of this minimal length is tied to the fact that strings can not probe distances shorter than the string length scale @xmath6 .
, other probes , such as @xmath7-branes @xcite , can probe scales smaller than @xmath6 .
in this situation , another type of uncertainty relation involving spatial and temporal coordinates is found to hold : @xmath8 @xcite . ]
thus , @xmath9 in ref .
@xcite , we determined the eigenvalues and eigenfunctions of the harmonic oscillator when the position and momentum obey eq .
( [ eq : com1 ] ) , and studied the possible constraint that can be placed on @xmath10 by precision measurements on electrons trapped in strong magnetic fields .
subsequently , in ref .
@xcite , we pointed out that eq .
( [ eq : com1 ] ) implies the finiteness of the cosmological constant and a modification of the blackbody radiation spectrum of the cosmic microwave background .
one important observation made in refs .
@xcite was that various observable effects of the minimal length uncertainty relation are non perturbative in the `` deformation parameter '' @xmath10 ( _ i.e. _ contain all orders in @xmath10 ) even though @xmath10 appears only to linear order in eqs .
( [ eq : com1 ] ) and ( [ uncert ] ) . in this paper
, we continue our investigation and consider the effects of the `` deformation '' of the canonical commutation relations on the orbits of classical particles in a central force potential .
we find that comparison with observation places a strong constraint on the size of the minimum length .
in @xmath7-dimensions , eq .
( [ eq : com1 ] ) is extended to the tensorial form @xcite : @xmath11 = i\hbar ( \delta_{ij } + \beta \hat{p}^2 \delta_{ij } + \beta ' \hat{p}_i \hat{p}_j ) \;. \label{eq : com2}\ ] ] if the components of the momentum @xmath12 are assumed to commute with each other , @xmath13 = 0\ ; , \label{eq : com3}\ ] ] then the commutation relations among the coordinates @xmath14 are almost uniquely determined by the jacobi identity ( up to possible extensions ) as @xmath15 = i\hbar\ , \frac{(2\beta-\beta ' ) + ( 2\beta+\beta')\beta\hat{p}^2 } { ( 1+\beta \hat{p}^2 ) } \left ( \hat{p}_i \hat{x}_j - \hat{p}_j \hat{x}_i \right)\;. \label{eq : com4}\ ] ] in the classical limit , the quantum mechanical commutator is replaced by the poisson bracket via @xmath16 \longrightarrow \{a , b\}\;. \label{climit}\ ] ] so the classical limits of eqs .
( [ eq : com2])([eq : com4 ] ) read @xmath17 we are keeping the parameters @xmath10 and @xmath18 fixed as @xmath19 , which in the string theory context corresponds to keeping the string _ momentum _ scale fixed while the string _ length _ scale is taken to zero .
note that for eq .
( [ climit ] ) to make sense , the poisson bracket must possess the same properties as the quantum mechanical commutator , namely , it must be anti - symmetric , bilinear , and satisfy the leibniz rules and the jacobi identity .
these requirements allows us to derive the general form of our poisson bracket for any functions of the coordinates and momenta as @xmath20 where repeated indices are summed .
in particular , we find that the time evolutions of the coordinates and momenta are governed by @xmath21 this `` deformed '' version of classical mechanics is not without its difficulties , the foremost being how one can construct `` canonical transformations '' which relate the dynamical variables at one length scale to those at another . for the minimal length to be a well defined length scale , all dynamical variables at all length scales must obey eq .
( [ eq : poisson1 ] ) . as a consequence , for instance
, one can not identify the position of a composite particle with the center of mass of its constituents .
in retrospect , it is not surprising that this difficulty would exist given the uv / ir mixing nature of eqs .
( [ eq : com2])([eq : com4 ] ) from which eqs .
( [ eq : poisson1 ] ) have been derived .
we merely point out this difficulty as a caveat and do not attempt to propose any solution in the current paper .
instead , we apply eq .
( [ eq : poisson3 ] ) to the motion of macroscopic objects and look for signatures of the deformation .
for the hamiltonian of a particle in a central force potential , @xmath22 the derivatives with respect to the coordinates and momenta are @xmath23 therefore , the time evolutions of the coordinates and momenta in this case are @xmath24\ , \frac{p_i}{m } - [ \ , ( 2\beta-\beta ' ) + ( 2\beta+\beta')\beta p^2 \ , ] \left ( \frac{1}{r } \frac{\partial v}{\partial r } \right ) l_{ij}\ , x_j \ ; , \cr \dot{p}_i & = & - [ \ , ( 1+\beta p^2)\ , x_i + \beta ' ( p\cdot x)\ , p_i \ , ] \left ( \frac{1}{r } \frac{\partial v}{\partial r } \right ) \;,\end{aligned}\ ] ] where @xmath25 the @xmath26 s defined here are the generators of rotation : @xmath27 for motion in a central force potential , the @xmath26 s are conserved due to rotational symmetry : @xmath28 so is @xmath29 the conservation of the @xmath26 s imply that the motion of the particle will be confined to a 2-dimensional plane spanned by the coordinate and momentum vectors at any point in time .
therefore , without loss of generality , we can assume that the motion is in the @xmath30-plane and @xmath31 then , the motion can be described by the time dependences of the distance from the origin @xmath32 , and the angle @xmath33 the equation of motion for @xmath32 is given by @xmath34\,p_r \ ; , \label{req}\end{aligned}\ ] ] where @xmath35 since the energy , @xmath36 , is also conserved , we can write the momentum squared as a function of @xmath32 via @xmath37\;. \label{eminusv}\ ] ] therefore , right hand side of eq .
( [ req ] ) can be written completely in terms of conserved quantities and functions of @xmath32 : @xmath38 \sqrt { 2 m ( e - v ) - { \frac{\displaystyle l^2 \left [ 1 + 2m\beta ( e - v ) \right]^2 } { \displaystyle r^2 } } } \;.
\label{req2}\ ] ] the equation of motion for the angle @xmath39 is @xmath40 ( 1+\beta p^2 ) + [ \,(2\beta-\beta ' ) + ( 2\beta+\beta')\beta p^2 \ , ] \left ( mr\frac{\partial v}{\partial r } \right ) \right\}\;. \label{phieq}\end{aligned}\ ] ] again , using eq . (
[ eminusv ] ) , the right hand side can be written in terms of conserved quantities and functions of @xmath32 only : @xmath41 \left[\ , 1 + 2m\beta ( e - v ) \ , \right ] \biggr . \cr & & \biggl .
+ \left[\ , ( 2\beta-\beta ' ) + 2m\beta(2\beta+\beta ' ) ( e - v ) \ , \right ] \left(\ , mr\frac{\partial v}{\partial r}\ , \right ) \biggr\}\;. \label{phieq2 } \end{aligned}\ ] ] from eqs .
( [ req2 ] ) and ( [ phieq2 ] ) , we find @xmath42 ^ 2 } { \displaystyle r^2 } } } } } \;. \label{dphidr}\ ] ] in principle , this equation can be integrated to obtain the @xmath39 dependence of @xmath32 .
we will solve eq .
( [ dphidr ] ) for the harmonic oscillator and coulomb potentials in the following two cases : 1 .
@xmath43 , @xmath44 , in which eq .
( [ dphidr ] ) simplifies to @xmath45 ^ 2 } { \displaystyle r^2 } } } } } \ ; , \label{dphidr2}\ ] ] and 2 .
@xmath46 , @xmath47 , in which eq .
( [ dphidr ] ) simplifies to @xmath48
we first consider the harmonic oscillator potential @xmath49 for the harmonic oscillator , eq .
( [ dphidr2 ] ) can be cast into the form @xmath50\ ; , \label{dphidrrhmo}\ ] ] where @xmath51 and @xmath52 @xmath53 are the turning points when @xmath43 and @xmath54 are the turning points when @xmath46 . when @xmath55 satisfies the condition @xmath56 it is possible to show that @xmath57 when @xmath55 exceeds the upper bound of the region eq . ( [ betaregion ] )
, no solution exists .
( [ dphidrrhmo ] ) can be integrated to yield @xmath58\;.\end{aligned}\ ] ] in particular , we find @xmath59 which shows that the orbit will not close on itself when @xmath43 .
it precesses by an angle of @xmath60 per revolution . for @xmath61 ,
the precession angle is @xmath62 in figure ( [ harmonic ] ) , we plot the trajectory of the motion for a representative set of parameters . for the harmonic oscillator , eq .
( [ dphidr3 ] ) can be cast into the form @xmath63 where @xmath64 note that the turning points , @xmath54 , do not depend on @xmath18 . in the limit @xmath65 , we have @xmath66 , and the equation for the @xmath67 case is recovered .
( [ dphidrrhmo2 ] ) can be integrated to yield @xmath68\ ; , \label{phirhmo2}\end{aligned}\ ] ] where @xmath69 note that @xmath70 in the limit @xmath65 . from eq .
( [ phirhmo2 ] ) , we find @xmath71 compared to the @xmath43 , @xmath44 case , eq .
( [ precessionhmo ] ) , the precession is in the opposite direction : for each revolution , the angle swept is smaller than @xmath72 by @xmath73 . for @xmath74 ,
the precession angle is @xmath75 in figure ( [ harmonic ] ) , we plot the trajectory of the motion for a representative set of parameters .
next , we consider the attractive coulomb potential @xmath76 for bound states , @xmath77 , eq .
( [ dphidr2 ] ) takes on the form @xmath78 where @xmath79 and @xmath80 \cr & & + \,i\,\varepsilon^{3/2}\,\frac{(r_{+}+r_{-})^2}{\sqrt{r_{+}r_{- } } } \left [ 1 + \frac{3}{8 } \left\ { 1 - \frac { 7\,(r_{+}+r_{-})^2 } { 8\,r_{+}r_{- } } \right\ } \varepsilon + \mathcal{o}(\varepsilon^2 ) \right]\;.\end{aligned}\ ] ] the exact forms of @xmath81 and @xmath53 are rather lengthy and non - illuminating , so we will not present them here .
( see appendix a. ) @xmath54 are the turning points when @xmath46 , and we can see that when @xmath82 , @xmath83 just as in the harmonic oscillator case . the condition that @xmath84 must satisfy for the solution to exist is @xmath85 eq .
( [ dphidrcoulomb ] ) can be integrated and the solution expressed exactly in terms of elliptic integrals .
( see appendix b. ) however , the exact expression is not particularly informative so we present the solution to linear order in @xmath10 , in which case we find @xmath86 \arcsin\left\ { \frac { ( r - r_{\min } ) r_{\max } - ( r_{\max}-r ) r_{\min } } { ( r_{\max}-r_{\min } ) r } \right\ } \cr & & + \frac { ( r_{\max } + r_{\min } ) } { r } \sqrt { \frac { ( r_{\max}-r)(r - r_{\min } ) } { r_{\max } r_{\min } } } \,\varepsilon + \mathcal{o}(\varepsilon^2)\;,\end{aligned}\ ] ] and @xmath87 \;.
\label{precessioncoulomb}\ ] ] note that , in contrast to the harmonic oscillator , the precession angle is negative : @xmath88 where @xmath89 is the eccentricity of the orbit .
this means that the perihelion of a planet in a gravitational coulomb potential will _ retard _ instead of advance .
in figure ( [ coulomb ] ) , we plot the trajectory of the motion for a representative set of parameters . for bound states , @xmath77 , eq .
( [ dphidr3 ] ) takes on the form @xmath90 where @xmath91 as in the harmonic oscillator case , the turning points @xmath54 do not depend on @xmath18 . in the limit @xmath65
, we have @xmath92 , and the equation for the @xmath67 case is recovered .
( [ dphidrcoulomb2 ] ) can be integrated to yield , @xmath93\ ; , \label{phircoulomb2}\end{aligned}\ ] ] where @xmath94 note that @xmath95 in the limit @xmath65 . from eq .
( [ phircoulomb2 ] ) , we find @xmath96\;. \label{precessioncoulomb2}\ ] ] as in the harmonic oscillator case , the precession angle is negative when @xmath18 is positive . for @xmath74 ,
the precession angle is @xmath97 in figure ( [ coulomb ] ) , we plot the trajectory of the motion for a representative set of parameters .
using our results , we can place constraints on @xmath10 and @xmath18 from the precession of the perihelion of mercury . according to ref .
@xcite , the observed advance of the perihelion of mercury that is unexplained by newtonian planetary perturbations or solar oblateness is @xmath98 this advance is usually explained by general relativity which predicts @xmath99 where @xmath100 is the schwarzschild radius of the sun , @xmath101 is the semi - major axis of the planet s orbit , @xmath89 is it s eccentricity , and we have defined @xmath102 the lengths @xmath103 and @xmath104 are the de broglie and compton wavelengths of the planet . for mercury , the parameters are @xcite @xmath105 note that the product @xmath106 is known to much better accuracy than newton s gravitational constant @xmath107 and the solar mass @xmath108 separately . using these parameters
we find @xmath109 and @xmath110 comparison of eqs .
( [ deltagr ] ) and ( [ deltaobserved ] ) yields @xmath111 which is consistent with zero .
as we can see , there is not much room left for possible extra contributions to the precession . from eq .
( [ delta1 ] ) and ( [ delta2 ] ) , the precession angle to linear order in @xmath10 and @xmath18 is @xmath112 the existence of a minimal length requires @xmath113 so we can assume that @xmath114 eq .
( [ deltaremainder ] ) places a lower bound on @xmath115 which at @xmath116 is @xmath117 thus , @xmath118 or @xmath119 note that this limit is 33 orders of magnitude below the planck length !
in this paper , we have considered the effects of the minimal length uncertainty relation on the classical orbits of particles in a central force potential .
comparison with the observed precession of the perihelion of mercury places a strong constraint on the value of the minimum length .
the minimal length uncertainty relation was implemented through the deformed commutation relation eq .
( [ eq : com2 ] ) .
note that even though @xmath10 and @xmath18 appear to only linear order on the right hand side of eq .
( [ eq : com2 ] ) , our expressions for the precession angle , eqs .
( [ precessionhmo ] ) , ( [ precessionhmo2 ] ) , ( [ precessioncoulomb ] ) , and ( [ precessioncoulomb2 ] ) , contain all orders in @xmath10 and @xmath18 . in that sense ,
our results are non - perturbative . on the other hand , the right hand side of eq .
( [ eq : com2 ] ) itself can be considered a linear approximation to a more general expression which leads to the minimal length uncertainty relation as discussed by kempf @xcite .
this suggests that our constraint , eq .
( [ superstrongconstraint ] ) , could be fairly robust .
all other possible implementation of the minimal length uncertainty relation can be expected to lead to the same precession of the perihelion as eq .
( [ eq : com2 ] ) to linear order in @xmath10 and @xmath18 , and result in the same constraint on the minimal length .
the analysis of this paper based on the deformed commutation relations can be viewed as providing a toy model for a full string theoretic consideration of the implications of the minimal length uncertainty relation .
the natural question to ask is whether our constraint , eq .
( [ superstrongconstraint ] ) , applies to string theory proper or not .
this is a difficult question to answer since the minimal length uncertainty relation is but one aspect of string theory , and it is not clear whether deforming the quantum mechanical commutation relations is the correct way to implement it . looking at previous works , we note that ref .
@xcite has discussed departures from general relativity as implied by string theory .
these were implied both by the string theoretic modification of einstein s equations @xcite @xmath120 as well as the crucial distinctions between particles and strings : strings as extended objects do not fall freely along geodesics . as fundamentally extended objects ( at least from the point of view of string perturbation theory )
they are subject to tidal forces .
this leads , for example , to an energy dependent deflection angle for the bending of light in clear distinction to general relativity in which the deflection angle is energy independent .
we have not included in our analysis any of these effects . in particular , we have not considered possible deviations in the background metric due to the extra terms in eq .
( [ stringeinstein ] ) .
though the corrections to particle trajectories due to such deviations are expected to be small , it may be worthwhile to study the problem in more detail in light of the strong constraint we have obtained for the minimal length .
we conclude by listing a few more caveats : even though our analysis is purely classical , the general formulation of classical systems which incorporates the classical limit of the minimal length uncertainty relation is not fully understood .
how one can define the `` canonical transformations '' which relate dynamical variables at different scales while preserving the poisson bracket remains an open problem .
also , the systems we considered have only a finite number of degrees of freedom .
it is not clear how to incorporate the effects of the classical limit of the minimal length uncertainty relation to field theory .
the classical limit of the minimal length uncertainty relation provides a natural generalization of the non commutative relation between spatial coordinates encountered in non commutative field theory @xcite .
what is not clear is whether the usual weyl wigner moyal technology @xcite could apply even in our more complicated set - up , thus providing a way to analyze systems with an infinite number of degrees of freedom .
we would like to thank yasushi nakajima , john simonetti , and joseph slawny for helpful discussions . this research is supported in part by a grant from the us department of energy , de fg0592er40709 .
the turning points for the coulomb potential , eq .
( [ coulomb ] ) , are provided by the real solutions to @xmath121 ^ 2 = 0\;. \label{a1}\ ] ] defining @xmath122 eq . ( [ a1 ] ) can be cast into the form @xmath123 since this is a quartic equation , the solutions can be obtained algebraically ( using _ mathematica _ ) and they are : @xmath124 where @xmath125^{1/3 } \cr & = & b + ( 3 a^2 - 2b)\,\varepsilon + \frac{a^2\sqrt { 3b(a^2 - 4b ) } } { b } \,\varepsilon^{3/2 } + \left ( 9a^2 + b - \frac{9a^4}{2b } \right)\varepsilon^2 + \cdots\;. \label{wxyzdef}\end{aligned}\ ] ] in the limit @xmath126 , we recover the turning points for the @xmath46 case : @xmath127
integration of eq . ( [ dphidrcoulomb ] ) yields @xmath128 where @xmath129 these integrals can be expressed in terms of the legendre
jacobi elliptic integrals @xcite : @xmath130 define @xmath131 with @xmath132 and @xmath133 where @xmath134 , @xmath135 , and @xmath136 are given in eq .
( [ wxyzdef ] ) .
the explicit expressions for the integrals are @xmath137 d. j. gross and p. f. mende , nucl .
b * 303 * , 407 ( 1988 ) ; d. j. gross and p. f. mende , phys .
b * 197 * , 129 ( 1987 ) ;
d. j. gross , phys . rev .
lett . *
60 * , 1229 ( 1988 ) .
d. amati , m. ciafaloni and g. veneziano , phys .
b * 216 * , 41 ( 1989 ) ; d. amati , m. ciafaloni and g. veneziano , int .
j. mod .
phys . a * 3 * , 1615 ( 1988 ) ; d. amati , m. ciafaloni and g. veneziano , phys .
b * 197 * , 81 ( 1987 ) ; k. konishi , g. paffuti and p. provero , phys .
b * 234 * , 276 ( 1990 ) .
e. witten , phys . today * 49 * ( 1997 ) 24 .
c. m. hull , jhep * 9807 * , 021 ( 1998 ) [ arxiv : hep - th/9806146 ] ; c. m. hull , jhep * 9811 * , 017 ( 1998 ) [ arxiv : hep - th/9807127 ] ; v. balasubramanian , p. horava and d. minic , jhep * 0105 * , 043 ( 2001 ) [ arxiv : hep - th/0103171 ] ; e. witten , hep - th/0106109 ; a. strominger , arxiv : hep - th/0106113
. t. banks , int .
j. mod .
a * 16 * ( 2001 ) 910 .
t. banks , arxiv : hep - th/0007146 . for other closely related attempts to understand the cosmological constant problem consult for example : t. banks , arxiv : hep - th/9601151 ; a. g. cohen , d. b. kaplan and a. e. nelson , phys .
lett . * 82 * , 4971 ( 1999 ) [ arxiv : hep - th/9803132 ] ; p. horava and d. minic , phys .
lett . * 85 * , 1610 ( 2000 ) [ arxiv : hep - th/0001145 ] ; n. arkani - hamed , s. dimopoulos , n. kaloper and r. sundrum , phys .
b * 480 * , 193 ( 2000 ) [ arxiv : hep - th/0001197 ] ; s. kachru , m. schulz and e. silverstein , phys .
d * 62 * , 045021 ( 2000 ) [ arxiv : hep - th/0001206 ] ; p. berglund , t. hubsch and d. minic , arxiv : hep - th/0112079 ; p. berglund , t. hubsch and d. minic , arxiv : hep - th/0201187 .
both observational constraints on and theoretical approaches to the cosmological constant are reviewed in s. m. carroll , arxiv : astro - ph/0004075 .
for some theoretical perspectives see s. weinberg , arxiv : astro - ph/0005265 ; rev .
phys . * 61 * , 1 ( 1989 ) , and e. witten , arxiv : hep - ph/0002297 .
a. kempf , phys .
d * 63 * , 083514 ( 2001 ) [ arxiv : astro - ph/0009209 ] ; a. kempf and j. c. niemeyer , phys .
d * 64 * , 103501 ( 2001 ) [ arxiv : astro - ph/0103225 ] ; r. easther , b. r. greene , w. h. kinney and g. shiu , phys .
d * 64 * , 103502 ( 2001 ) [ arxiv : hep - th/0104102 ] ; r. easther , b. r. greene , w. h. kinney and g. shiu , arxiv : hep - th/0110226 ; l. mersini , m. bastero - gil and p. kanti , phys .
d * 64 * , 043508 ( 2001 ) [ arxiv : hep - ph/0101210 ] ; m. bastero - gil and l. mersini , phys .
d * 65 * , 023502 ( 2002 ) [ arxiv : astro - ph/0107256 ] ; m. bastero - gil , p. h. frampton and l. mersini , arxiv : hep - th/0110167 ; n. kaloper , m. kleban , a. e. lawrence and s. shenker , arxiv : hep - th/0201158 .
t. yoneya , int .
j. mod .
phys . a * 16 * , 945 ( 2001 ) [ arxiv : hep - th/0010172 ] ; t. yoneya , arxiv : hep - th/9707002 ; m. li and t. yoneya , phys . rev . lett .
* 78 * , 1219 ( 1997 ) [ arxiv : hep - th/9611072 ] ; m. li and t. yoneya , arxiv : hep - th/9806240 ; h. awata , m. li , d. minic and t. yoneya , jhep * 0102 * , 013 ( 2001 ) [ arxiv : hep - th/9906248 ] ; d. minic , phys .
b * 442 * , 102 ( 1998 ) [ arxiv : hep - th/9808035 ] ; d. minic and c. h. tze , arxiv : hep - th/0202173 . _
allen s astrophysical quantities _ , 4th edition , edited by a. n. cox ( springer - verlag , 2000 ) ; jpl solar system dynamics group website , http://ssd.jpl.nasa.gov/phys_props_planets.html ; d. e. groom _ et al . _
[ particle data group collaboration ] , eur .
j. c * 15 * , 1 ( 2000 ) . , @xmath138 , @xmath44 ( left ) , and @xmath46 , @xmath139 ( right ) .
the length scale is in units of @xmath140 .
the red line indicates the orbit when @xmath67 .
the motion is counter clockwise along the trajectory starting from the perihelion on the positive @xmath141 axis .
25 complete revolutions are shown .
for both cases , the trajectory is precessing clockwise.,title="fig : " ] , @xmath138 , @xmath44 ( left ) , and @xmath46 , @xmath139 ( right ) .
the length scale is in units of @xmath140 .
the red line indicates the orbit when @xmath67 .
the motion is counter clockwise along the trajectory starting from the perihelion on the positive @xmath141 axis .
25 complete revolutions are shown .
for both cases , the trajectory is precessing clockwise.,title="fig : " ] , @xmath138 , @xmath44 ( left ) , and @xmath46 , @xmath139 ( right ) .
the length scale is in units of @xmath140 .
the red line indicates the orbit when @xmath67 .
the motion is counter clockwise along the trajectory starting from the perihelion on the positive @xmath141 axis .
25 complete revolutions are shown .
for both cases , the trajectory is precessing clockwise.,title="fig : " ] , @xmath138 , @xmath44 ( left ) , and @xmath46 , @xmath139 ( right ) .
the length scale is in units of @xmath140 .
the red line indicates the orbit when @xmath67 .
the motion is counter clockwise along the trajectory starting from the perihelion on the positive @xmath141 axis .
25 complete revolutions are shown .
for both cases , the trajectory is precessing clockwise.,title="fig : " ] | we continue our investigation of the phenomenological implications of the `` deformed '' commutation relations @xmath0=i\hbar[(1+\beta\hat{p}^2)\delta_{ij } + \beta'\hat{p}_i\hat{p}_j]$ ] .
these commutation relations are motivated by the fact that they lead to the minimal length uncertainty relation which appears in perturbative string theory . in this paper
, we consider the effects of the deformation on the classical orbits of particles in a central force potential .
comparison with observation places severe constraints on the value of the minimum length . |
the description of the algebraic topology of the moduli space @xmath2 of compact complex algebraic curves has long been a tantalizing problem .
the idea of `` stable cohomology '' of @xmath2 as the genus @xmath5 , brought in by j. l. harer and d. mumford , suggested a more graspable object to study as a first step .
mumford s conjecture @xcite , stating that the stable cohomology is the polynomial algebra on the so - called mumford - morita - miller classes , has become a new tantalizing problem since then .
recently m. pikaart @xcite has proven that the hodge structure on the stable cohomologyis pure , thus providing more evidence to the conjecture , which easily implies purity .
our purpose in this note is to show that the rational homotopy type of the moduli space @xmath2 stabilizes as @xmath5 and prove that the moduli space is formal in the stable limit . if the stable cohomology of the moduli space were just known to be the cohomology of a certain space @xmath6 , then a theorem of e. miller and s. morita @xcite asserting that @xmath7 is a free graded commutative algebra would immediately imply the formality of @xmath6
thus , the essential result of this paper is showing that the limit of the rational homotopy types of @xmath2 as @xmath8 exists .
after this is done , the formality of the limit is automatic .
here are few words about the structure of the paper .
sections 1 and 2 are dedicated to the description of the ( nonstable ) rational homotopy type of the moduli space @xmath2 .
section [ infty ] , where the stable rational homotopy type of @xmath2 is studied , is independent of the first two sections .
it may be worth noting how this paper relates to jim stasheff s work .
stasheff polyhedra @xmath9 are known for many remarkable properties .
one of them , that @xmath9 is a connected component of the real compactification of the moduli space of real projective lines with @xmath10 punctures , was pointed out by m. kontsevich in @xcite .
thus , stasheff polyhedra are real analogues of the compactified moduli spaces @xmath11 . or , perhaps , the moduli spaces @xmath11 are complex analogues of stasheff polyhedra , the chicken and the egg problem .
the topology of @xmath9 is trivial ( @xmath9 is contractible ) , and the topology of @xmath2 and @xmath11 is just the opposite .
it is rather the combinatorics of stasheff polyhedra which makes them very useful in the topology of loop spaces @xcite .
the combinatorics of the genus zero spaces @xmath12 has similarly proven to be useful in studying the topology of double loop spaces @xcite and the algebraic structure of 2d quantum field theory @xcite . but what is the topology of these complex analogues of stasheff polyhedra , especially for a high genus ?
we also implicitly use a result of s. halperin and stasheff @xcite throughout the paper : formality over any field of characteristic zero implies that over the rationals .
i am very grateful to pierre deligne , dick hain , eduard looijenga , and martin pikaart for helpful discussions .
i would like to thank the institute for advanced study in princeton and the research institute for mathematical sciences in kyoto for their hospitality during the summer 1997 , when the essential part of the work on this paper was done .
here we recall j. w. morgan s description @xcite of a model of the rational homotopy type of the complement @xmath13 of a normal crossings divisor @xmath14 in a compact complex manifold @xmath15 .
the rational homotopy type will be understood in the sense of d. sullivan @xcite , see also p. a. griffiths and morgan @xcite and d. lehmann @xcite .
it determines the usual rational homotopy type of a topological space if it is simply connected . in general
it determines the rational nilpotent completion thereof .
the rational homotopy type of @xmath13 is determined by the differential graded ( dg ) commutative algebra that is nothing but the first term @xmath16 of the spectral sequence associated with the weight filtration on the log - complex @xmath17 of the smooth @xmath18-forms on @xmath13 with logarithmic singularities along @xmath19 .
otherwise , one can describe the same dg algebra as the second term of the leray spectral sequence of inclusion @xmath20 . in any case , the model of @xmath13 is given by the dg algebra @xmath21 , where @xmath22 the multiplication structure is given by @xmath23 for @xmath24 and @xmath25 , where @xmath26 is the sign of the shuffle putting the set @xmath27 in an increasing order , assuming each of the subsets @xmath28 and @xmath29 to be already in an increasing order .
note that this multiplication law makes @xmath30 into a bigraded commutative algebra .
finally , the differential @xmath31 can be described as @xmath32 where @xmath33 , @xmath34 , @xmath35 is the natural embedding @xmath36 , and @xmath37 is the gysin map .
p. deligne @xcite proved that the spectral sequence @xmath38 degenerates at @xmath39 , that is , the cohomology of the dg algebra @xmath21 is equal to @xmath40 .
the very dg algebra @xmath21 describes the rational homotopy type of @xmath13 , according to morgan s theorem @xcite .
results of the previous section apply to the moduli space @xmath2 , which is the complement of a normal crossing divisor in the deligne - knudsen - mumford compactification @xmath11 .
the problem that the space in question is a stack rather than variety does not arise , because we work with the * complex coefficients * , as we will assume throughout the paper . from the combinatorics of deligne - knudsen - mumford s construction
, we can say more specifically , cf .
@xcite , that @xmath41 the summation running over all stable labeled @xmath4-graphs @xmath42 of genus @xmath43 and @xmath44 vertices . here
we refer to _ graphs _ of the following kind .
each graph is connected and has _
@xmath4 enumerated exterior edges _ , edges which are incident with only one vertex of the graph .
each vertex @xmath45 of the graph is _ labeled _ by a nonnegative integer @xmath46 , called the genus of a vertex .
the _ stability _
condition means that any vertex @xmath45 labeled by @xmath47 should be incident with at least one edge ( i.e. , be at least of valence one ) and each vertex @xmath45 with @xmath48 should be at least of valence three .
the _ genus _ @xmath49 of a graph @xmath42 is given by the formula @xmath50 , where @xmath51 is the first betti number of the graph .
number @xmath52 _ is the valence of the vertex @xmath45 and @xmath53 is the _ automorphism group of a graph @xmath42 _ ( bijections on vertices and edges , preserving the exterior edges , the labels of vertices and the incidence relation ) .
the subscript @xmath54 in the formula refers to taking the homogeneous component of this degree .
the differential @xmath31 is induced by contracting interior edges in @xmath42 , which corresponds to replacing a neighborhood of a double point on a curve by a cylinder .
our goal here is to look at the stable ( as @xmath5 ) cohomology and rational homotopy type of the moduli space @xmath2 .
the following result of pikaart @xcite gives a certain clue on what is going on within a `` stable range '' .
[ pikaart ] the restriction mapping @xmath55 \to \linebreak[0 ] h^k({\mathcal{m}_{g , n}})$ ] is surjective for @xmath56 . for @xmath57 , the cohomology of @xmath58 is nonzero only for @xmath59 , see figure @xmath60 .
= 1.5 in by our construction of the spectral sequence , the natural composite mapping @xmath61 is the same as the restriction mapping @xmath62 .
therefore , the @xmath63 columns of @xmath58 do not contribute to the cohomology of the dg algebra @xmath64 in the stable range .
the stable limit in cohomology of the moduli spaces @xmath2 is achieved , roughly speaking , by gluing more and more handles to the complex curve .
this yields an inductive system ( of isomorphisms ) on the level of cohomology .
but since there is no natural mapping between the moduli spaces of different genera , one can not speak of a limiting rational homotopy type .
the question of taking the limit of the dg algebras @xmath64 of section [ mgn ] for @xmath2 may not be so obviously resolved , either , because these algebras are constructed out of the cohomology of @xmath65 , @xmath66 running between 0 and @xmath3 .
when @xmath67 , @xmath66 does not , and on top of that , taking the stable limit of cohomology of @xmath11 requires a finer tuning , cf .
pikaart @xcite .
our plan here is to show that a @xmath1-minimal model of @xmath2 is independent of @xmath1 as long as @xmath0 . in particular , the limit of @xmath1-minimal models exists and may be called a `` @xmath1-minimal model of @xmath6 '' , continuing the abuse of notation adopted for cohomology . since a minimal model of a space
may be obtained as a union of @xmath1-minimal models , we call this union a `` minimal model of @xmath6 '' . each of these @xmath1-minimal models is @xmath1-formal in a natural sense , see below , and the formality of @xmath6 follows .
first of all , we recall basic notions on @xmath1-minimal models , see @xcite . from now on we will assume that our spaces and algebras are connected and simply connected , i.e. , their @xmath68 and @xmath69 . in application to moduli spaces , this is the case as long as @xmath70 , see harer @xcite .
a dg algebra @xmath71 is called _ minimal _ if it is free as a dg commutative algebra , @xmath72 , and @xmath73 , where @xmath74 .
a _ minimal model _ of a dg algebra @xmath64 is a minimal dg algebra @xmath71 along with a _ quasi - isomorphism _
@xmath75 , a morphism of dg algebras inducing an isomorphism on cohomology .
every dg algebra @xmath64 has a minimal model , unique up to an isomorphism , which is in its turn unique up to homotopy .
let @xmath76 be an integer .
a @xmath1-_minimal model _ of a dg algebra @xmath64 is a minimal algebra @xmath77 generated by elements in degrees @xmath78 along with a morphism @xmath79 inducing an isomorphism on cohomology in degrees @xmath78 and an injection in degree @xmath80 .
a @xmath1-minimal model is unique up to an isomorphism uniquely defined up to homotopy .
if one has an increasing sequence of embeddings @xmath81 together with morphisms @xmath79 compatible with each other , so that @xmath77 is a @xmath1-minimal model of @xmath64 , then the union @xmath82 along with the natural morphism @xmath75 is a minimal model of @xmath64 .
we will call a dg algebra @xmath64 ( or a space whose rational homotopy type we consider ) _ formal _ if a minimal model @xmath71 of @xmath64 is isomorphic to a minimal model of its cohomology @xmath83 taken with the zero differential .
this is equivalent to saying that there is a quasi - isomorphism @xmath84 . in this case
the rational homotopy type of @xmath64 ( or the space ) is determined by its rational cohomology ring .
formality implies that all massey products are zero .
if the cohomology algebra @xmath83 is free as a graded commutative algebra , then @xmath64 is formal ; see proposition [ k - formal ] for a @xmath1-version of this statement .
another example of a formal space is any compact khler manifold , a famous result of griffiths , deligne , morgan , and sullivan @xcite . we will similarly call a dg algebra @xmath64 or a space @xmath1-_formal _ , if @xmath1-minimal models of @xmath64 and its cohomology are isomorphic .
we also need to prove the following proposition establishing a particular case of @xmath1-formality .
we say that a graded commutative algebra @xmath85 is _ @xmath1-free _ if there exists a graded vector space @xmath86 and a mapping @xmath87 of graded vector spaces defining a morphism @xmath88 of graded algebras , where @xmath89 is the free dg commutative algebra on @xmath90 , which is an isomorphism in degrees @xmath78 and an injection in degree @xmath80 .
[ k - formal ] suppose that the cohomology @xmath83 of a dg algebra @xmath64 is @xmath1-free , based on a graded vector space @xmath90 .
then @xmath89 is a @xmath1-minimal model of @xmath64 , and @xmath64 is @xmath1-formal .
if @xmath83 is @xmath1-free , then there exists a linear injection @xmath91 .
pick a linear mapping @xmath92 which takes each element @xmath45 of @xmath90 to a cocycle representing the cohomology class @xmath93 .
then the natural morphism of graded commutative algebras @xmath94 obviously respects the differentials and satisfies the axioms of a @xmath1-minimal model of @xmath64 . by assumption , @xmath89 is at the same time a @xmath1-minimal model of @xmath83 , whence @xmath1-formality of @xmath64 . 1 .
the moduli space @xmath2 is @xmath1-formal for @xmath0 .
the subalgebra @xmath95 of @xmath7 generated in degrees @xmath78 is a @xmath1-minimal model of @xmath2 for @xmath0 .
a @xmath1-minimal model of @xmath2 is independent of @xmath3 as long as @xmath0
. we will call it a @xmath1-_minimal model of _ @xmath6 .
the @xmath1-minimal models of @xmath6 form am increasing sequence of embeddings .
the union , a _ minimal model of _
@xmath6 , is isomorphic to its cohomology @xmath96 . in particular
, @xmath6 is formal .
the stable cohomology @xmath96 is a free graded commutative algebra , _
i.e. _ , isomorphic to @xmath89 for a graded vector space @xmath90 , according to miller - morita s theorem @xcite for @xmath97 and looijenga s handling @xcite of the @xmath98 case .
moreover , @xmath99 and @xmath100 for @xmath70 , see harer @xcite . given a nonnegative integer @xmath1 ,
the groups @xmath101 are known to stabilize as soon as @xmath102 ; this is the harer - ivanov stability theorem @xcite .
therefore , @xmath103 is @xmath1-free for @xmath0 : the mapping @xmath104 makes @xmath105 a @xmath1-free graded algebra .
proposition [ k - formal ] then implies that @xmath2 is @xmath1-formal for @xmath0 , @xmath106 being a @xmath1-minimal model of it . 2 and 3 .
notice that since @xmath96 is free , the subalgebra @xmath107 generated in degrees @xmath78 can be identified with @xmath106 , which we have just seen to be a @xmath1-minimal model of @xmath2 for @xmath0 . in view of this result ,
mumford s conjecture @xcite , if true , implies the following refinement : _ the polynomial algebra on the mumford - morita - miller classes @xmath109 , @xmath110 and the first chern classes @xmath111 of the `` tangent at the @xmath112th puncture '' bundles , @xmath113 , with a zero differential is the stable minimal model of the moduli space @xmath2 as @xmath5_. nikolai v. ivanov . on the homology stability for teichmller modular groups : closed surfaces and twisted coefficients . in _ mapping class groups and moduli spaces of riemann surfaces ( gttingen , 1991/seattle , wa , 1991 ) _ , volume 150 of _ contemp .
_ , pages 149194 .
soc . , providence , ri , 1993 .
t. kimura , j. stasheff , and a. a. voronov .
homology of moduli of curves and commutative homotopy algebras . in _ the gelfand mathematical seminars ,
19931995 _ , gelfand math .
, pages 151170 .
birkhuser boston , boston , ma , 1996 . `
alg - geom/9502006 ` .
t. kimura , a. a. voronov , and g. j. zuckerman .
homotopy gerstenhaber algebras and topological field theory . in _ operads : proceedings of renaissance conferences ( hartford , ct / luminy , 1995 ) _ , volume 202 of _ contemp .
_ , pages 305333 , providence , ri , 1997 .
q - alg/9602009 ` . | we show that for @xmath0 the @xmath1-rational homotopy type of the moduli space @xmath2 of algebraic curves of genus @xmath3 with @xmath4 punctures is independent of @xmath3 , and the space @xmath2 is @xmath1-formal .
this implies the existence of a limiting rational homotopy type of @xmath2 as @xmath5 and the formality of it . |
neutron - star low - mass x - ray binaries ( ns - lmxbs ) with weak magnetic fields show a wide variety in x - ray spectral and variability properties . based on the correlated behavior of these properties ,
various ns - lmxb subclasses have been recognized ; the atoll sources , the sco - like z sources , and the cyg - like z sources @xcite .
observations of the z - source transient xtej1701462 have shown that these sub - classes are linked through mass accretion rate , with the atolls having the lowest mass accretion rates and the cyg - like z sources having the highest ones @xcite .
the differences in mass accretion rate between the sub - classes is likely the result of differences in system parameters , such as orbital period and properties of the donor star .
however , for many ns - lmxbs these parameters are not well known , making it difficult to look for systematic differences in the system parameters between the ns - lmxb sub - classes .
summary of the values of the spectral index @xmath4 that result from fits to the fire spectrum of gx9 + 1 .
the spectrum was dereddened using different assumptions about @xmath3 , @xmath5 ( ps=@xcite , go=@xcite ) , and the nir extinction law ( n09=@xcite , ccm89=@xcite . ) as discussed in sect .
[ sec_nhd ] , we consider the lower @xmath3 values less likely as they imply a rather small distance and low x - ray luminosity for gx9 + 1 . if the actual @xmath3 towards gx9 + 1 is indeed on the high end of the range considered , the positive slope of the nir spectrum of gx9 + 1 points at thermal emission . given the uncertainties in the extinction law and @xmath5 ratios , we refrain from trying to distinguish between viscous heating and heating by x - ray reprocessing .
the positive slope is inconsistent with a dominant contribution from jet synchrotron emission in this wavelength region . in this respect ,
gx9 + 1 is similar to the other two bright atolls and most z sources that have been studied in the nir , which also have nir spectra with positive slopes . for gx9 + 1 and the other bright atolls ,
there is no indication for the presence of a strong jet given that these sources have not been detected at radio wavelengths @xcite .
this is not surprising as jets are commonly associated with ns - lmxbs that are in relatively hard x - ray spectral states .
the bright atolls , on the contrary , are typically found in an x - ray soft state .
therefore , we can also turn the argument around : the assumption that there is no jet in gx9 + 1 , implies that @xmath4 must be positive .
this can be used as another argument in favor for the higher @xmath3 values towards the system .
if the thermal nir emission is mainly coming from the outer disk , where x - ray heating dominates over viscous heating , there should be a connection between the @xmath0-band luminosity , the x - ray luminosity , and the orbital period ( which sets the size of the disk ) .
the existence of such a correlation has been demonstrated by @xcite , analogously to the work by @xcite for the case of the optical emission of x - ray binaries .
we have used their empirical formula to constrain the orbital period of gx9 + 1 based on its x - ray luminosity and estimated absolute @xmath0 magnitude .
we assume that @xmath670% of the x - ray luminosity in the 0.1218 kev band is emitted in the 210 kev band , for which revnivtsev et al .
derived their relation .
the resulting range is not very restrictive : for @xmath7 kpc and @xmath8 @xmath9 we find @xmath10 h , whereas @xmath11 kpc and @xmath12 @xmath9 give @xmath13 h. again , the choice of @xmath5 ratio has a small ( @xmath145% ) effect .
follow - up photometric monitoring of the nir counterpart may reveal the actual orbital period of gx9 + 1 .
the long - term variation in the x - ray light curve that is clearly visible in fig .
[ fig_longterm ] is unlikely to be related to the orbit but could reflect very - low frequency noise or solar - like magnetic - activity cycles @xcite the authors would like to thank j. fridriksson , r. remillard , p. sullivan , and m. matejek for obtaining part of the observations . we acknowledge the use of data products from vvv survey observations made with the vista telescope at the eso paranal observatory under program i d 179.b-2002 . , e. , mellier , y. , radovich , m. , et al .
2002 , in astronomical society of the pacific conference series , vol .
281 , astronomical data analysis software and systems xi , ed .
d. a. bohlender , d. durand , & t. h. handley , 228 , p. , persson , s. e. , murphy , d. c. , et al .
2004 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 5492 , society of photo - optical instrumentation engineers ( spie ) conference series , ed .
a. f. m. moorwood & m. iye , 16531660 | we have determined an improved position for the luminous persistent neutron - star low - mass x - ray binary and atoll source gx9 + 1 from archival _ chandra x - ray observatory _ data .
the new position significantly differs from a previously published _
chandra _ position for this source . based on the revised x - ray position we have identified a new near - infrared ( nir ) counterpart to gx9 + 1 in @xmath0-band images obtained with the panic and fourstar cameras on the magellan baade telescope .
nir spectra of this @xmath1 mag star taken with the fire spectrograph on the baade telescope show a strong br @xmath2 emission line , which is a clear signature that we discovered the true nir counterpart to gx9 + 1 .
the mass donor in gx9 + 1 can not be a late - type giant , as such a star would be brighter than the estimated absolute @xmath0 magnitude of the nir counterpart .
the slope of the dereddened nir spectrum is poorly constrained due to uncertainties in the column density @xmath3 and nir extinction . considering the source s distance and x - ray luminosity
, we argue that @xmath3 likely lies near the high end of the previously suggested range .
if this is indeed the case , the nir spectrum is consistent with thermal emission from a heated accretion disk , possibly with a contribution from the secondary . in this respect ,
gx9 + 1 is similar to other bright atolls and the z sources whose nir spectra do not show the slope that is expected for a dominant contribution from optically thin synchrotron emission from the inner regions of a jet .
this paper includes data gathered with the 6.5 meter magellan telescopes located at las campanas observatory , chile . ] |
the well known artin representation of the braid group @xmath0 may be used to calculate the group of a knot . applying fox free differential calculus to this representation
, we can derive the burau representation .
its irreducible part may be used to calculate the alexander polynomial of a knot . in @xcite the author
extend the artin and the burau representation to a representation of the baez - birman singular braid monoid @xmath1 .
a monoid is said to be linear if it is isomorphic to a submonoid of matrices @xmath2 for some natural number @xmath3 and some field @xmath4 . in @xcite
the authors showed the faithfulness of the two dimensional extended burau representation of @xmath5 , therefore this monoid is linear .
it is natural then to search for a faithful representation of the surface singular braid monoid @xmath6 defined in @xcite , where the author classified knotted surfaces in @xmath7 that have surface singular braid index equal to one or two , and also showed that there exist infinitely many surface - link types that are closures of elements from @xmath8 .
we will show in this paper that any representation of @xmath6 , for any @xmath9 , to the multiplicative monoid of all @xmath10 or @xmath11 matrices with entries in a given field , is not faithful .
we will also derive new presentations for the surface singular braid monoid , one with reduced the number of defining relations , and the other with reduced the number of its defining non - classical generators .
an embedding ( or its image ) of a closed ( i.e. compact , without boundary ) surface into @xmath7 is called the _ knotted surface _ ( or _ surface - link _ ) .
two knotted surfaces are _ equivalent _ ( or have the same _ type _ ) , if there exists an orientation preserving homeomorphism of the four - space @xmath7 to itself ( or equivalently auto - homeomorphism of the four - sphere @xmath12 ) , mapping one of those surfaces onto the other .
we will work in the standard smooth category .
let @xmath13 denote @xmath14 for @xmath15 .
it is well known ( @xcite , @xcite , @xcite ) that for any surface - link @xmath16 , there exists a surface - link @xmath17 satisfying the following : @xmath17 is equivalent to @xmath16 and has only finitely many morse s critical points , all maximal points of @xmath17 lie in @xmath18 , all minimal points of @xmath17 lie in @xmath19 , all saddle points of @xmath17 lie in @xmath20 . the zero section @xmath21 of the surface f gives us then a 4-valent graph .
we assign to each vertex a _ marker _ that informs us about one of the two possible types of saddle points ( see fig . [ pic03 ] ) depending on the shape of @xmath22 or @xmath23 for a small real number @xmath24 .
the resulting graph is called a _ marked graph_. making now a projection in general position of this graph to @xmath25 and assigning types of classical crossings between regular arcs , we obtain a _ marked graph diagram_. for a marked graph diagram @xmath26 , we denote by @xmath27 and @xmath28 the diagrams obtained from @xmath26 by smoothing every vertex as presented in fig .
[ pic03 ] for @xmath29 and @xmath30 , respectively .
./pic03(4 cm ) any two marked graph diagrams representing the same type of knotted surface are related by a finite sequence of yoshikawa local moves presented in fig .
[ pic04 ] ( and an isotopy of the diagram in @xmath25 ) .
./pic04(14.5 cm )
we can present every marked diagram of a surface - link in a _ braid form _ defined as the geometric closure of a singular braid with markers .
we have the monoid @xmath31 that corresponds to marked graph diagrams in braid form on @xmath32 strands . for @xmath33
this monoid is trivial with one element , let us assume that @xmath34 .
elements of @xmath31 called _ surface singular braids _ , are generated by four types of elements @xmath35 for @xmath36 , where the correspondence of types of crossings and types of markers between @xmath37-th and @xmath38-th strand ( in the horizontal position , numbered from the top to the bottom ) is presented in fig .
[ pic05 ] .
./pic05(5.5 cm ) let @xmath39 , @xmath34 and @xmath40 such that @xmath41 , moreover let @xmath42 .
monoid @xmath31 is a subject to the following relations .
1 . @xmath43 , 2 .
@xmath44 for @xmath45 , 3 .
@xmath46 , 4 .
@xmath47 , 5 .
@xmath48 , 6 .
@xmath49 for @xmath50 , 7 .
@xmath51 for @xmath50 , 8 .
@xmath52 , 9 .
@xmath53 , 10 .
@xmath54 , 11 .
@xmath55 .
we will indicate our closure of a marked diagram in a braid form by adding square brackets around its words and adding lower index after it , saying how many strands we are joining .
let us further denote by @xmath56 a subset of @xmath31 containing only those elements @xmath57 , that @xmath58_m)$ ] and @xmath59_m)$ ] are diagrams of trivial classical links .
we define the additional relations on closed braids .
1 . @xmath60_n=\left[s_nx_i\right]_n$ ] for @xmath61 and @xmath62 and @xmath63 2 .
@xmath64_n=\left[s_nx_n\right]_{n+1}$ ] for @xmath61 and @xmath65 [ thm4 ] making change in a closed braid word formulation of a knotted surface by using one of relations from @xmath66 or @xmath67 , we receive a formula of a knotted surface of the same type . [ prop2 ] monoid @xmath31 for @xmath39 and @xmath34 is generated by @xmath68 for @xmath69 , @xmath42 and is a subject to the following relations : @xmath70 some relations from ( a2)-(a4 ) that includes @xmath71 are known , from classical singular braid theory , to follow from ( r1)-(r9 ) .
the remaining relations are either the same or derived as follows .
@xmath72 @xmath73 @xmath74 sometimes ( for computational reasons ) we want to have less generators and therefore the following presentation is useful . [ prop3 ] monoid @xmath6 for @xmath75 and @xmath76
is generated by @xmath77 and @xmath78 for @xmath79 and is a subject to the following relations : @xmath80 set @xmath81 and introduce @xmath82 for @xmath83 by relations ( r6 ) and ( r7 ) .
it is sufficient to derive specific ( different ) relations from the sets ( r1)-(r16 ) , ( m1)-(m16 ) , that have not been already derived in the proof of prop .
2.2 in @xcite ( when @xmath84 is replaced here either by @xmath85 or @xmath86 , and @xmath87 is replaced by @xmath88 ) . for @xmath89 relations ( m10 ) , ( m14)-(m16 )
are easily equivalent to ( r10 ) , ( r14)-(r16 ) respectively .
we now derive relations ( r10)-(r16 ) for @xmath83 .
relation ( r10 ) follows from ( m10 ) and the following inductive step @xmath90 relation ( r11 ) follows from ( m11 ) and the following inductive step @xmath91 relation ( r12 ) follows from ( m12 ) and the following inductive step @xmath92 relation ( r13 ) follows from ( m13 ) by the similar argument as in relation ( r12 ) .
+ relation ( r14 ) follows from ( m14 ) and the following inductive step @xmath93 relation ( r15 ) follows from ( m15 ) and the following inductive step @xmath94 relation ( r16 ) follows from ( m16 ) and the following inductive step @xmath95 [ prop1 ] we have the following all ( un)knotted surfaces whose surface singular braids can be defined with two strands @xmath96_2,$ ] @xmath97_2,$ ] @xmath98_2,$ ] @xmath99_2,$ ] @xmath100_2,$ ] @xmath101_2.$ ] the @xmath3-twist - spun surface - knot of the classical rational link @xmath102 $ ] in conway notation encodes as @xmath103)=[ac_2^{k_{2m+1}}c_1^{-k_{2m}}\cdots c_2^{k_1}bc_2^{-k_1}c_1^{k_2}\cdots c_2^{-k_{2m+1}}(c_1c_2c_1)^{2n}]_3.\ ] ] some of knotted surfaces in yoshikawa s table are included in the above case as follows : @xmath104)$ ] , @xmath105)$ ] , @xmath106)$ ] , @xmath107)$ ] , @xmath108)$ ] , @xmath109)$ ] . these and algebraic formulations of other knotted surfaces in yoshikawa s table are summarized in table [ tab01 ] . @xmath110_1\\ \hline 2_1^{1 } , \text { unknotted } \mathbb{t}^2 & [ ab]_2\\ \hline 2_1^{-1 } , \text { unknotted } \mathbb{p}^2_+ & [ ac_1]_2\\ \hline \text{unknotted } \mathbb{p}^2_- & [ bc_1]_2\\ \hline \text{unknotted } \mathbb{kb}^2 & [ abc_1]_2\\ \hline 7_1^{0,-2 } & [ abc_2^{-1}c_1^{-2}c_2^{-1}c_1^{-1}]_3\\ \hline 10_2^{0,1 } & [ ab(c_2c_1 ^ 2c_2)^2]_3\\ \hline \tau^n(\text{rational link } c[k_1,k_2,\ldots , k_{2m+1 } ] ) & [ ac_2^{k_{2m+1}}c_1^{-k_{2m}}\cdots c_2^{k_1}bc_2^{-k_1}c_1^{k_2}\cdots c_2^{-k_{2m+1}}(c_1c_2c_1)^{2n}]_3\\ \hline 8_1^{1,1 } , \text { spun surface of hopf link } & [ ( abc_2^{-1}c_3^{-1}c_1c_2)^2]_4\\ \hline 8_1^{-1,-1 } & [ bc_2^{-1}c_1^{-1}c_2c_1 ^
2c_3^{-1}c_2bc_2^{-1}c_1^{-1}c_2^{-1}c_1 ^ 2c_3^{-1}c_2]_4\\ \hline 9_1^{0,1 } & [ abc_2^{-1}c_3^{-1}c_2 ^ 2c_3^{-1}c_2c_1 ^ 2c_2]_4\\ \hline 9_1^{1,-2 } & [ ( abc_2^{-1}c_3^{-1}c_1c_2)^2c_1^{-1}]_4\\ \hline 10_1^{1 } , \text { spun torus of the trefoil } & [ ac_2c_3c_1c_2bc_2^{-1}c_1^{-1}c_3^{-1}c_2 ^ 3c_3c_1c_2ac_2^{-1}c_1^{-1}c_3^{-1}c_2^{-1}bc_2^{-3}]_4\\ \hline 10_1^{1,1 } & [ ac_2^{-1}c_1^{-1}c_3c_2^{-1}bc_2^{-1}c_1 ^ 2c_2ac_2^{-1}c_1^{-1}c_3^{-1}c_2bc_2 ^ 2]_4\\ \hline 10_1^{0,0,1 } & [ abc_3c_2^{-1}c_1^{-2}c_2^{-1}c_3^{-1}c_2c_1 ^
2c_2]_4\\ \hline 10_1^{0,-2 } & [ abc_2^{-1}c_1^{-1}c_2^{-1}c_1^{-1}c_3c_2^{-2}c_3]_4\\ \hline 10_2^{0,-2 } & [ ac_2^{-1}c_3c_2c_1^{-1}c_2 ^ 2bc_2^{-1}c_1^{-1}c_3^{-1}c_2^{-1}c_1c_2^{-1}]_4\\ \hline 10_1^{-1,-1 } & [ ac_2^{-1}c_1^{-1}c_3^{-1}c_2bc_2c_1^{-1}c_2c_3c_2^{-1}c_1c_2^{-1}c_1 ^ 2c_2]_4\\ \hline 10_1^{-2,-2 } & [ ( abc_2^{-1}c_3^{-1}c_1c_2c_1^{-1})^2]_4\\ \hline 9_1 & [ ac_2^{-1}c_1^{-1}c_3^{-1}c_2^{-1}c_4^{-1}c_3^{-2}c_2c_1^{-1}c_3c_2c_3c_4c_2c_3c_1c_2bc_2^{-1}\cdot\\ & \cdot c_1^{-1}c_3^{-1}c_2^{-1}c_4^{-1}c_5^{-1}c_6^{-1}c_4^{-1}c_5c_7^{-1}c_6^{-1}c_4^{-1}c_3^{-1}c_4c_2^{-1}c_3^{-1}\cdot\\ & \cdot c_1c_2^{-1}c_4c_5c_4c_3c_4c_5^{-1}c_6c_5^{-1}c_4^{-1}c_7c_6c_5c_3c_4c_2c_3c_1c_2]_8\\ \hline \end{array}$ ] [ prop5 ] in monoid @xmath6 for @xmath75 and @xmath76 the following relations hold . @xmath111 for elements of the monoid @xmath6 it follows that ( see thm .
[ thm4 ] and prop .
[ prop1 ] ) : @xmath112_n=\mathbb{kb}^2\sqcup\underbrace{\mathbb{s}^2\sqcup\ldots\sqcup\mathbb{s}^2}_{n-2}\not=\mathbb{t}^2\sqcup\underbrace{\mathbb{s}^2\sqcup\ldots\sqcup\mathbb{s}^2}_{n-2}=[ab]_n,\ ] ] @xmath113_n=[bc_1^{-1}]_n=\mathbb{p}^2_+\sqcup\underbrace{\mathbb{s}^2\sqcup\ldots\sqcup\mathbb{s}^2}_{n-2}\not=\mathbb{p}^2_-\sqcup\underbrace{\mathbb{s}^2\sqcup\ldots\sqcup\mathbb{s}^2}_{n-2}=[bc_1]_n=[ac_1^{-1}]_n.\ ] ] this implies relations ( e1)-(e3 ) .
consider now the spun @xmath114-knot of the trefoil , it is the well known nontrivial @xmath114-knot , as its group is isomorphic to the group of the classical trefoil .
it follows from prop .
[ prop1 ] that this knotted @xmath114-sphere can be presented as @xmath115_3 $ ] , we also have trivial @xmath114-sphere @xmath116_3 $ ] ( see @xcite for the proof ) , therefore we have @xmath117_n\not=[ac_2^{-3}bc_2^{3}]_n\not=[ac_2^{-3}bc_2^{3}(c_1c_2c_1)^{2}]_n\not=[abc_1c_2]_n=\mathbb{kb}^2\sqcup\bigsqcup_{n-3}\mathbb{s}^2.\ ] ] this , together with relations ( m1 ) , ( m3)-(m7 ) , ( m14 ) , ( m15 ) implies relations ( e4)-(e7 ) .
let @xmath4 throughout this paper denote a field . by a _ representation _ of a monoid @xmath26 of dimension @xmath3 over @xmath4 we mean a homomorphism @xmath118 of @xmath26 into the multiplicative monoid of @xmath2 of all @xmath119 matrices with entries in @xmath4 .
if @xmath118 is injective then the representation is said to be _
faithful_. denote @xmath120 and @xmath121 the identity matrix and the zero matrix of size @xmath122 respectively .
[ prop6 ] for @xmath123 and any faithful representation @xmath124 , there is a faithful representation @xmath125 such that : @xmath126 let us recall the following property of idempotent matrix .
[ lem1 ] if a matrix @xmath127 with entries in a field @xmath4 satisfies @xmath128 then it is diagonalizable and all its eigenvalues are either @xmath129 or @xmath130 . consider @xmath127 as an endomorphism operator on a vector space @xmath131 . take any nonzero vector @xmath132 , then there exists @xmath133 such that @xmath134 , from the idempotency relation @xmath128 we have @xmath135 which yields @xmath136 , so we have @xmath137 , therefore @xmath127 is diagonalizable .
if @xmath138 is its eigenvalue then there exists nonzero vector @xmath133 such that @xmath139 .
we must have then that @xmath140 , and because @xmath4 is a field , this implies @xmath141 .
the monoid @xmath142 of @xmath143 matrices over @xmath4 can be identified with @xmath144 , the monoid of endomorphisms of a vector space @xmath131 over @xmath4 of finite dimension @xmath32 . applying lem .
[ lem1 ] for @xmath145 , we can conclude that there exists a matrix @xmath146 such that @xmath147 , where @xmath148 .
we define a new representation by setting @xmath149 for any @xmath150 , and now its injectivity follows immediately from injectivity of @xmath151 .
it proves ( p1 ) beside cases @xmath152 , @xmath153 which will be excluded later . from prop .
[ prop5 ] we have @xmath154 , @xmath155 and @xmath156 , and together with relations ( m1 ) and ( m12)-(m14 ) we moreover have @xmath157 , @xmath158 , @xmath159 , hence from injectivity of @xmath118 we have cases ( p2 ) , ( p3 ) , ( p6 ) , ( p11 ) , ( p12 ) and remaining cases @xmath160 from ( p1 ) . relations ( p4 ) and ( p5 ) follow from ( m14 ) together with ( p11 ) and ( p12 ) respectively .
remaining relations ( p7)-(p10 ) follow directly from ( e4)-(e7 ) .
[ prop4 ] if a representation @xmath125 for @xmath123 satisfies @xmath161 or @xmath162 then @xmath118 is not faithful . from the symmetric role of @xmath85 and @xmath86 in @xmath6 , we can assume that @xmath161 .
denote @xmath163 and @xmath164 . from relation ( p1 ) of prop .
[ prop6 ] we can assume that @xmath165 , then from @xmath166 it follows that @xmath167 and @xmath168 .
this implies that @xmath169 , and combining it with @xmath170 gives us relation @xmath171 that contradicts relation ( p4 ) of prop .
[ prop6 ] . from prop .
[ prop6 ] and prop .
[ prop4 ] we immediately have the following .
no representation @xmath172 for @xmath173 is faithful .
a faithful representation @xmath118 of the monoid @xmath174 can be defined ( in a field of characteristic zero ) as follows @xmath175 [ thm1 ] no representation @xmath176 for @xmath9 is faithful .
assume the contrary , that @xmath118 is faithful and denote @xmath177 , for @xmath178 .
from relations ( m11 ) and ( m13 ) of prop .
[ prop3 ] we have @xmath170 , and @xmath166 . from prop .
[ prop6 ] we can assume that @xmath179 , @xmath180 , @xmath181 , @xmath182 and that @xmath183 .
let @xmath184 . then from relation
@xmath166 it follows that @xmath185 for some matrix @xmath186 . from relation
@xmath170 it follows that @xmath187 and @xmath188 .
if @xmath189 then @xmath190 , a contradiction , so it follows that @xmath191 , and combining it with @xmath192 gives moreover @xmath193 .
consider now @xmath194 , from relation ( m4 ) it follows that @xmath195 for some matrix @xmath196 .
non - invertability of matrix @xmath197 together with relation @xmath182 implies that @xmath198 and @xmath199 , additionally from relations ( m1 ) , ( m5 ) and ( m14 ) we have @xmath200 , @xmath201 and @xmath202 .
consider the following two main cases .
* case a * , assume @xmath203 .
then by @xmath193 we must have that @xmath204 . from the symmetric role of @xmath205 and @xmath206 , without loss of generality ,
assume @xmath207 .
then from @xmath187 and @xmath198 we obtain that @xmath197 is one of the two possible forms : @xmath208 or @xmath209 . from @xmath201
it follows that @xmath210 , and relation @xmath202 together with @xmath199 yields that @xmath16 is one of the two possible forms @xmath211 or @xmath212 one for each mentioned type of @xmath197 respectively .
additionally from ( m1 ) and ( p11 ) we have that @xmath213 , @xmath214 in the first and @xmath215 , @xmath216 in the second case respectively . from ( m1 ) and ( p12 )
we moreover have @xmath217 and @xmath218 in both cases .
* case b * , assume @xmath219 . then from relations ( m1 ) , ( m4 ) , ( m5 ) , ( m11 ) , ( m13 ) , ( m14 ) , ( p2 ) , ( p4)-(p6 ) , ( p11 ) and ( p12 ) it follows that @xmath220 are in the form : @xmath221 for @xmath222 , @xmath223 , @xmath224 , @xmath225 and @xmath218 .
introducing now a matrix @xmath226 , and making elementary computations ( that can be found in appendix of the arxiv source file of this article s preprint version ) , we can show that in both of the above cases , relations ( m1 ) , ( m3 ) , ( m6 ) , ( m7 ) , ( m10 ) , ( m15 ) , ( p7 ) , ( p8 ) and ( p10 ) form a self - contradictory set .
the author s research was partially supported by grant bw - ug for young researchers , and warsaw center of mathematics and computer science
. 99 o.t .
dasbach , b. gemein , _ a faithful representation of the singular braid monoid on three strands _ , in knots in in hellas 98 , ( world scientific publishing , 2000 ) 48 - 58
. b. gemein , _ representations of the singular braid monoid and group invariants of singular knots _ , topology and its applications * 114 * ( 2001 ) 117 - 140 .
m. jablonowski , _ on a surface singular braid monoid _ , topology and its applications
* 160 * ( 2013 ) 1773 - 1780 .
y. joung , j. kim and s. y. lee , _ on generating sets of yoshikawa moves for marked graph diagrams of surface - links _ , _
j. knot theory ramifications _ * 24 * ( 2015 ) , no . 4 , 1550018 . s. kamada , non - orientable surfaces in 4-space , _
osaka j. math . _
* 26 * ( 1989 ) , 367 - 385 .
a. kawauchi , t. shibuya , and s. suzuki , _ descriptions on surfaces in four - space , i ; normal forms _ , math .
knotes kobe univ .
* 10 * ( 1982 ) 72 - 125 . c. kearton , v. kurlin , _
all 2-dimensional links in 4-space live inside a universal 3-dimensional polyhedron _ , algebraic and geometric topology * 8*(3 ) ( 2008 ) 1223 - 1247 .
lomonaco , jr .
, _ the homotopy groups of knots i. how to compute the algebraic 2-type _ , pacific j. math . * 95 * ( 1981 ) 349 - 390 .
swenton , _ on a calculus for 2-knots and surfaces in 4-space _ , j. knot theory ramifications * 10 * ( 2001 ) 1133 - 1141 .
k. yoshikawa , _ an enumeration of surfaces in four - space _ , osaka j. math .
* 31 * ( 1994 ) 497 - 522 .
e. c. zeeman , _ twisting spun knots _ , trans .
* 115 * ( 1965 ) 417 - 495 . | the surface singular braid monoid corresponds to marked graph diagrams , of knotted surfaces in braid form . in a quest to resolve linearity problem for this monoid
, we will show that if it is defined on at least two or at least three strands , then its two or respectively three dimensional representations , are not faithful .
we will also derive new presentations for the surface singular braid monoid , one with reduced the number of defining relations , and the other with reduced the number of its singular generators .
we include surface singular braid formulations of all knotted surfaces in yoshikawa s table . |
in recent years , nanophotonics has attracted considerable attention as one of the emerging fields of modern science and technology .
active devices , especially coherent sources with high efficiency , are of great importance for the development of nanophotonics .
the amplification of the evanescent field and strong confinement of light in a subwavelength scale are crucial for active nanophotonic devices to achieve superior performance .
surface plasmon polaritons ( spps ) provide a feasible route for manipulating light beyond the diffraction limit . in 2003 , bergman and stockman proposed surface plasmon amplification by stimulated emission of radiation ( spaser)@xcite . a spaser referring to the accumulation of a large number of spps in a single mode assisted by a gain medium is precisely the plasmonic counterpart of the nano - laser . in principle ,
one of the most important applications of spp amplification is plasmon lasing@xcite . as energy dissipation in metals significantly reduces the lifetime of spps in the infrared and visible frequencies ,
the gain of the active medium must prevail over the dissipation of metals in order to sustain the survival of a spaser .
many efforts related to spp amplification have been devoted to loss compensation in metal - dielectric structures@xcite .
the first experimental verification of the spaser used gold nano - particles encapsulated in dye - doped silica shells , which support a collective spp mode with a high quality factor ( q - factor)@xcite .
planar design is preferred for the realization of spasing as well as plasmon lasing for the sake of fabrication convenience and device integration .
however , as a finite - sized planar structure is usually incapable of strong light confinement , a high - gain medium is required in order to compensate for the radiation and damping losses from the planar structure .
one solution to this difficulty is to utilize trapped mode of a plasmonic meta - surface@xcite .
a split ring resonator with weak asymmetry ( asr ) possesses a leaky mode that is poorly coupled with free - space photons . a plasmonic metamaterial slab composed of an array of such asrs will facilitate spp amplification and lasing with low - gain material when the leaky mode is excited .
subsequent experimental investigation showed that the loss from the nano - sized metallic asrs could be compensated for by the gain of quantum dots@xcite .
it is worth noting that a magnetic meta - surface can support spatially coherent spps which are of high q - factor as well .
the subtle concept , in contrast to the trapped mode which is derived from dark mode of a single resonator , comes from the spps in harmonic mode@xcite .
these spatially coherent spps are actually the leaky waves of a slab waveguide which can not be determined by a single resonator .
they must be in harmonic mode to sustain a leaky wave character , giving rise to a high q - factor which can be of one or two orders larger as compared to that of the spps in fundamental mode .
in this paper , we theoretically propose that such spatially coherent spps present a practical solution for low - threshold plasmon lasing in both the near - infrared and visible regimes .
the gain coefficient @xmath3 of the active medium for threshold lasing can be tuned down by structural parameters to a large extent , even down to values much smaller than those of semiconductor quantum wells .
two model samples are presented with structural parameters that are feasible for practical fabrication .
the threshold gain coefficient @xmath465 @xmath0 at 193.5 thz ( 1.55@xmath1 m ) is much smaller than those of other lasing systems reported in the same infrared frequency region@xcite .
the low threshold arises from the high q - factor of the spps in harmonic mode and from the strong confinement of the local field in the active layer , both of which are not easy to establish in a planar solid - state system .
the high directionality of the lasing beam with the full width at half maximum ( fwhm ) of @xmath5 in e - plane ( @xmath6 ) and @xmath7 in h - plane ( @xmath8 ) is verified by finite - difference - in - time - domain ( fdtd ) simulations , where @xmath9 and @xmath10 are the polar angle and azimuthal angle .
our planar lasing system is schematically illustrated in the inset of fig .
lying on the @xmath11 plane , the slab comprises an upper layer of a metallic lamellar grating with a thickness of @xmath12 , a metallic ground plane , and an active layer with a thickness of @xmath13 sandwiched between them .
the metallic strips with a width of @xmath14 are separated by a small air gap @xmath15 , giving rise to a period of @xmath16 for the lamellar grating .
the geometric parameters of our model in the infrared region are @xmath17 m , @xmath18 m , @xmath19 m , @xmath20 m , and @xmath21 m . the spp resonances on the slab can be characterized by the amplification of reflection coefficients .
the metals in our system are assumed to be silver with permittivity defined by a drude model@xmath22 , @xmath23 where @xmath24 @xmath25thz , @xmath26thz . and
the dielectric layer is assumed to be active with a complex permittivity @xmath27 , in which the imaginary part @xmath28 is positive .
we perform a rigorous coupled - wave analysis@xmath29 to calculate the coefficient @xmath30 of the @xmath31 order of the reflected bloch wave components , in which @xmath9 and @xmath32 denote the incident angle and the frequency of plane wave incidence .
the incident plane wave is transverse magnetic ( tm ) polarized with the magnetic field parallel to the @xmath33 direction and in - plane wave vector @xmath34 , where @xmath35 is the wave vector in a vacuum .
the gain strength of the active layer is defined by a coefficient , @xmath36 . in the calculations ,
different values of gain strength are taken into account by varying the imaginary part @xmath28 of the complex permittivity while keeping the real part fixed at @xmath37 .
figure 1 shows the calculated reflection spectra in the 0@xmath38 order under an incident angle of @xmath39 @xmath40 and an imaginary part of the permittivity of the active layer of @xmath41 @xmath42 .
when the gain strength is strong ( @xmath43 ) , there exists an amplification peak fixed at 46.3 thz ( @xmath44 m ) that is almost independent of the incident angle .
in contrast , when the gain strength is weak ( @xmath45 ) , the amplification peak at 46.3 thz disappears , and a new peak emerges at 193.5 thz ( @xmath46 m ) under normal incidence . under oblique incidence , the emerging peak splits into two peaks with varying frequencies that are sensitive to the incident angle ; see , for example , the dotted line in fig .
1(b ) for the two peaks at 166.1 thz and 211.9 thz with @xmath47 . the angle - dependent amplification peaks are obviously different from the angle - independent peak in regards to the physics of their origin .
their time - reversed counterparts are the angle - dependent and angle - independent absorption peaks that we have investigated before@xmath48 .
an angle - independent peak , requiring a sufficiently large gain / absorption coefficient of the dielectric layer , comes from the excitation of a localized spp resonance in a fundamental mode that is dominated by the 0@xmath38 order guided bloch mode in the dielectric .
the angle - dependent feature comes from the spatially coherent spp resonance in a harmonic mode that is dictated primarily by the @xmath491@xmath50 orders of the guided bloch mode in the dielectric .
the reflection amplification of the two resonant peaks as a function of the gain strength in the active layer is plotted in fig .
it should be noted that a rather low gain strength @xmath51@xmath0is sufficient to compensate for the loss of metals and to sustain the amplification of the guided waves in the active layer as well as the reflectance amplification at 193.5 thz .
the amplified reflectance increases exponentially with the increase of @xmath3 , until a maximal amplification of about @xmath52 is reached at an optimal value of @xmath53@xmath0 . in contrast , we see from fig .
2 that the reflectance amplification at 46.3 thz requires much stronger gain strength , which is unattainable for most of the existing active media at this frequency .
additional calculations indicate that the gain strength required for the reflection amplification at 193.5 thz can be reduced further , provided that a smaller air gap or a thinner dielectric layer which matches the gain strength is adopted .
the property provides us with a good opportunity for finding a balance between ease of fabrication and attainable gain strength of the active layer .
usually thermal conductivity of active medium becomes larger with respect to higher temperature .
a weaker gain strength required for threshold lasing also implies that we can find an appropriate structure for plasmon lasing that is robust against thermal noise , or intuitively that the thermal noise is somewhat suppressed by spatially coherent spps .
this is advantageous for room - temperature plasmon lasing as one goal being pursued . at this stage
, it is imperative for us to apply the spatially coherent spp modes for low - threshold plasmon lasing as such a spp mode with a long lifetime , which is compatible to the structural parameters , will create extensive adaptability in material selection and feasible way for practical implementation of plasmon lasing .
now we investigate the threshold condition of our lasing system by using the realistic material parameters . a widely used gain medium , ytterbium - erbium co - doped material er : yb : ycob is adopted as the active medium for lasing around @xmath54m@xcite@xmath55 its level assignment is shown in fig4 ( a ) .
the population density on every level evolves according to the dynamic rate equations@xcite , and can be solved for steady - state conditions:@xmath56 @xmath57 @xmath58 @xmath59 @xmath60 @xmath61 and @xmath62 are the population density and radiative lifetime of corresponding levels .
@xmath63 are the branching ratio for transitions from the @xmath64 to @xmath65 levels .
@xmath66 is the pump rate from 1y to 2y level . @xmath67 and @xmath68 are the forward and backward energy transfer coefficient corresponding to 2y@xmath691e@xmath703e@xmath691y and backwards .
@xmath71 is the coefficient of the cooperative upconversion 2e@xmath692e@xmath724e@xmath691e .
@xmath73 is the stimulated - emission coefficient .
@xmath74 is the inverted population difference between 1e and 2e level .
@xmath75 is the population density of the lasing mode .
considering the conservation of population density , we have @xmath76 and @xmath77 .
the threshold lasing can be evaluated under steady - state condition as @xmath78 , where @xmath79 is the q - factor of the lasing mode .
we note that the threshold inversion density @xmath80 can be estimated by @xmath79 ; and a higher q - factor as a result of lower radiation loss and damping loss of the lasing mode will give rise to lower threshold doping concentrations or pumping power dominated by @xmath80 . the q - factor of spp mode can be extracted from the fdtd numerical simulations by assuming that the dielectric is passive and lossless with a dielectric constant equal to 1.7 , the real part of that of er : yb : ycob . a one - way tm polarized plane wave with a gaussian distribution in time domain is normally incident on the model slab .
the incident plane is @xmath81 m away from the top surface of the model slab .
a probing plane is positioned @xmath82 m behind the incident plane to monitor the reflected wave fields .
figure 3(a ) illustrates the time - domain spectroscopy of the magnetic local field @xmath83 at the probing plane .
the frequency spectra in figs .
3(b ) and 3(c ) present the fourier transformed data of the time - domain spectroscopy in different time spans @xmath84ps and @xmath85ps .
a broad peak and a narrow one , centered at 45.0thz and 192.6thz respectively , are observed in both figures .
more calculations in time domain indicate that the decay rate of the broad one is much quicker than that of the narrow one .
these two peaks correspond to the localized spp resonance in fundamental mode at 46.3thz and the spatially coherent spp resonance in harmonic mode at 193.5thz , as shown in figs . 1 and 2 .
q - factor of a spp resonance can be extracted from the fwhm in the frequency spectra .
a low q - factor of only 4.2 and a high q - factor of 94.2 are derived at 45.0thz and 192.6thz in the frequency spectra .
the q - factor can also be estimated with the exponential decay rate of the mode in time domain , as @xmath86 , where @xmath87 is the lifetime of the mode . in our case , the life time of each mode is , @xmath88ps and @xmath89ps , respectively , reproducing exactly the same values of q - factor at the two resonance frequencies as the former method .
we note that the q - factor of spatially coherent spp mode is of one order larger than that of localized spp mode . and a higher q - factor will be advantageous in the field amplification in gain medium and the stimulated emission radiations . by applying the condition of threshold lasing at 192.6thz to the rate equations , a threshold doping concentration @xmath90 ions
/ m@xmath91 of er@xmath92 is obtained , which falls inside the typical concentration range of er@xmath92 ions in er : yb : ycob . in our calculations ,
the yb@xmath92 concentration is assumed to be fixed at @xmath93 ions / m@xmath91 .
it is not likely to utilize a fundamental spp mode of the meta - surface to realize plasmon lasing due to the low q - factor of 4.2 , as the required er@xmath92 concentration @xmath94 ions / m@xmath91 is hard to achieve in er : yb : ycob .
more calculations show that it is not very easy for the q - factor of the local spp resonance in fundamental mode to exceed a value of 10 by tuning structural parameters .
now we perform fdtd simulations to dynamically demonstrate a lasing example of our model slab@xmath95 .
periodic boundary is adopted in our simulations . the yb@xmath92 concentration and the er@xmath92 concentration
are chosen at 5.0@xmath9610@xmath97 irons / m@xmath91 and 4.0@xmath9610@xmath98 irons / m@xmath91 , respectively .
calculations are performed by assuming a pumping power at @xmath99w which is beyond the threshold @xmath100w _ _ .
the magnetic field @xmath83 pattern of the cavity and its emission field distribution at a typical time ( @xmath101ps ) are shown in fig .
4(b ) shows the time dependence of the averaged @xmath83 over all the spatial points in @xmath102 plane covering the unit cell .
the lasing at 192.6thz of a real structure is also considered . in the fdtd simulations ,
the lateral size of the model slab is 20 periods along @xmath103 direction ( the direction of the grating ) @xmath104 , and the length of the stripe along @xmath33 direction is @xmath105 .
the angular dependence of far - field patterns are calculated at the steady - state .
we see from fig . 5 that most of the energy of the lasing beam is confined in a small solid angle .
the angle - dependent power is normalized to the maximum at the normal direction .
the fwhm in the e - plane ( @xmath6 ) is @xmath106 , and that in the h - plane ( @xmath8 ) is @xmath7 .
the highly directional lasing beam shall be attributed to the excitation of a spatially coherent spp mode on the slab , as such an spp mode will inherently establish phase correlations among the unit cells on the meta - surface due to the dominant @xmath491@xmath50 orders of guided bloch modes in the active layer .
our lasing system can also operate at visible wavelength with an appropriate design .
figure 6 presents an example of reflectance amplification at 474 thz ( @xmath107 nm ) .
a rather low gain coefficient of @xmath108@xmath0 is enough for the compensation of metal dissipation .
even the optimal gain coefficient , @xmath109@xmath0 for the maximum reflectance , falls in the range of the gain coefficients of the most commonly used active media in this frequency regime , such as organic dye molecules , semiconductor quantum wells , and nano - crystal quantum dots .
the structural parameters are as follows : grating period @xmath110 nm , metallic strip width @xmath111 nm , air gap width @xmath112 nm , grating thickness @xmath113 nm , and dielectric layer thickness @xmath114 nm .
this calculation demonstrates that the spatial coherence of a harmonic magnetic spp resonance can be harvested for plasmon lasing in the visible region .
in summary , we demonstrate that spatially coherent spp modes in harmonic mode on a planar plasmonic meta - surface can be utilized for low - threshold directional plasmon lasing . as both the q - factor of the spatially coherent spp modes and the confinement of the local field are dictated primarily by structural parameters , the spp amplification can be triggered at a rather low threshold gain strength attainable with most of the active media .
meanwhile our examples indicate that the planar low - threshold plasmon lasing is not difficult to be realized with current fabrication techniques , even in the visible region .
our calculations show that using the harmonic spp modes is crux of the matter for achieving low - threshold lasing on a magnetic meta - surface rather than using the fundamental ones .
this work was supported by nsfc ( no . 10974144 , 60674778 ) , cnkbrsf ( grant no .
2011cb922001 ) , hk rgc grant 600308 , the national 863 program of china ( no .
2006aa03z407 ) , ncet ( 07 - 0621),stcsm , and shedf ( no . 06sg24 ) .
xy jiang and w li acknowledge nsfc ( grant no . 11004212 , 10704080 , 60877067 and 60938004 ) stcsm ( grant no .
08dj1400303 ) and snsfc(grant no .
11zr1443800 ) . | we theoretically propose directional , low - threshold plasmon lasing in both the near - infrared and visible wavelengths by utilizing spatially coherent surface plasmon polaritons on a meta - surface .
the gain strength required for threshold lasing can be tuned down to a large extent through compatible structural parameters .
our calculations show that no more than 65 @xmath0 at 193.5 thz ( 1.55 @xmath1 m ) or 267 @xmath2at 474thz ( 0.633 @xmath1 m ) of gain coefficient is sufficient to compensate for the dissipation of metal films for threshold lasing ; these values are smaller than any reported studies at the same frequencies .
these findings present a planar solid - state route for plasmon lasing that is highly efficient and spatially coherent . |
the role of kinetic effects in stimulated raman backscatter ( srbs ) is of much current interest .
srbs is the three - wave parametric coupling of a pump light wave ( the laser , labeled mode 0 ) to a counter - propagating daughter light wave ( mode 1 ) and co - propagating electron plasma wave ( epw , mode 2 ) , and satisfies the resonance conditions @xmath0 and @xmath1 .
it may remove a substantial amount of energy from a high - intensity laser propagating through a plasma and create energetic electrons @xcite .
srbs and other laser - plasma interactions must be sufficiently controlled for laser - driven inertial fusion to succeed . in the high - temperature plasmas expected on ignition experiments such as the national ignition facility ( nif )
@xcite and laser mgajoule ( lmj ) @xcite , linear theory of parametric instabilities predicts srbs to have a weak spatial gain rate due to the large @xmath2 ( @xmath3^{1/2}$ ] is the electron debye length ) and resulting heavy landau damping .
however , recent experiments @xcite and kinetic simulations @xcite have shown srbs reflectivities well above coupled - mode convective gain levels . this kinetic inflation or enhancement of srbs
is widely attributed to electron trapping in the epw potential well , which flattens the electron distribution function near the epw phase velocity @xmath4 @xcite . as has long been known
, trapping reduces the epw damping rate @xcite and induces an amplitude - dependent frequency downshift @xcite .
srbs saturation results from the competition of nonlinear effects including the damping reduction , frequency shift , pump depletion , and trapped particle instability @xcite , and possibly spatio - temporal chaos after ion modes build up near the laser entrance @xcite , among others .
a theory for the onset of kinetic inflation and the resulting time - averaged reflectivity is not yet in hand , but would be of great value for designing laser fusion implosions .
this current work is aimed at better understanding kinetic effects such as electron trapping in srbs for time scales and frequencies such that ion dynamics can be ignored . to this end , we have performed 1-d eulerian vlasov - maxwell simulations of srbs using the elvis code @xcite .
a rapid increase of reflectivity above the convective value is seen as we vary parameters such as the pump intensity .
the system does not reach a temporal steady state when kinetic enhancement occurs , but instead the reflectivity comes in sub - picosecond bursts ( as seen in , e.g. , ref .
@xcite ) . besides enhancing srbs , trapping also changes the electrostatic modes present .
electrostatic spectra from our simulations reveal the epw dispersion curve splits into an upper branch and several beam acoustic modes ( bams ) @xcite , and that srbs involves one of the latter .
the srbs daughter bam activity occurs along the streak that phase - matches for pump decay , which extends down in frequency from the linear phase - matching point with a slope equal to the daughter light wave s group velocity @xmath5 . this behavior has been seen in thomson scattering spectra from recent trident experiments @xcite and pic simulations @xcite . a low - amplitude acoustic wave ( meaning that @xmath6 , not an ion acoustic wave ) , which we call the electron acoustic wave ( eaw ) , also appears once inflation develops in our simulations .
moreover , weak reflected light that phase - matches for scattering off this mode is also seen , which we call electron acoustic scatter ( eas ) .
this process is reminiscent of what was called stimulated electron acoustic scatter ( seas ) and observed in trident single hot spot experiments @xcite ( although at the most intense pumps , the experimental seas levels relative to srbs are much stronger than in our simulations ) . to our knowledge , this paper is the first to analyze in detail eaws and eas arising from the srbs dynamics , and not from manually - distorted distributions or other strong seeding .
seas has also been reported in pic simulations of plasmas overdense to srbs and at relativistic pump intensities @xcite , and may have been seen in underdense vlasov simulations @xcite .
most of the eaw energy in our simulations is much too low - frequency to match eas and must be produced independently .
we propose the three - wave coupling of two bams and an eaw , which can be called beam acoustic decay ( bad ) , as the eaw excitation mechanism .
smaller - amplitude eaws at higher frequency are subsequently generated by harmonic coupling .
the pump laser then scatters off these fluctuations , referred to below as electron acoustic thomson scatter ( eats ) . the phase - matching needed for bad and eats
is seen both in electrostatic @xmath7 spectra and bispectral analysis .
we study the electrostatic mode structure by solving the linear dispersion relation for simulation distributions projected onto a gauss - hermite basis .
this yields the bams and eaw seen in our simulation spectra , and shows a damping reduction and frequency downshift ( compared to the epw for a maxwellian ) of the bam involved in srbs .
the bams are linearly unstable - in the absence of parametric coupling to light waves - for some wavenumber range . in addition
, we find an eaw root with heavy damping .
this linear eaw differs from the nonlinear eaw due to trapping discussed by others @xcite , which unlike ours is undamped and cuts off at @xmath8 ( in the limit of zero field amplitude ) .
the paper is organized as follows : section [ sec : model ] presents the model equations and geometry used in the elvis code .
the dependence of srbs on pump strength for simulations of trident single hot spot conditions is discussed in sec .
[ sec : tridgen ] .
kinetically - inflated srbs , and the related beam acoustic modes and electron acoustic scatter , are carefully studied in sec .
[ sec : trid2e15 ] for a pump intensity of 2 pw/@xmath9 .
section [ sec : hohl ] demonstrates that similar physics occurs in higher temperature , higher density plasmas found in hohlraum fills .
we discuss our results ( including experimental relevance ) and conclude in sec .
[ sec : conc ] .
the appendix details the gauss - hermite mode - finding method .
we use the eulerian vlasov - maxwell solver elvis @xcite , which resembles the code presented in @xcite .
our code implements the simplest model which allows for electron kinetic effects in raman scattering .
the geometry is one - dimensional , with all spatial variation and wavevectors in the @xmath10 direction .
we describe the particle species by the nonrelativistic vlasov equation in @xmath10 and keep the ions fixed in this paper .
the light waves are linearly polarized in @xmath11 , and the particles constitute a cold collisionless fluid in this direction ( the light waves are undamped ) .
the geometry is illustrated in fig .
[ fig : model_geom ] .
1-d geometry of the elvis code . ]
the governing equations are : @xmath12 f_{s } \\ & = & \nu_{ks}\left ( x\right ) \left ( n_{s}\hat{f}_{0s } - f_{s } \right ) , \\ \partial _ { x}e_{x } & = & \frac{e}{\varepsilon _ { 0}}\sum_{s}z_{s}n_{s } ,
\\ m_{s}\partial _
{ t}v_{ys } & = & ez_{s}e_{y } , \\ \left ( \partial _ { t}\pm c\partial _ { x}\right ) e^{\pm } & = & -\frac{e}{\varepsilon _ { 0}}\sum_{s}z_{s}n_{s}v_{ys } , \\ e^{\pm } & \equiv & e_{y}\pm cb_{z}. \end{aligned}\ ] ] @xmath13 , @xmath14 , and @xmath15 are the mass , charge , and number density of species @xmath16 ; @xmath17 for electrons , @xmath18 , and @xmath19 is the positron charge .
a number - conserving krook relaxation operator is included , with relaxation rate @xmath20 and equilibrium distribution @xmath21 set to the initial maxwellian ( @xmath22 ) .
we use a large @xmath23 ( @xmath24 ) at the edges of the finite length density profile to absorb plasma waves generated by srbs and to reduce fluctuations generated as the electron density adjusts to a sheath - like pattern .
a nonzero central value of @xmath25 can mimic transverse sideloss from a laser speckle , although @xmath26 outside the damping regions in this paper .
the vlasov equation for the distribution function @xmath27 is solved on a fixed phase - space grid via operator splitting @xcite .
the spatial and velocity advections are performed via a semi - lagrangian method ( shift along characteristics ) with cubic spline interpolation .
the @xmath10 shift is periodic with @xmath28 , although the dynamics in the central region are effectively finite in @xmath10 due to the damping regions and sheaths at the edges ( subscripts @xmath29 and @xmath30 denote quantities at the left and right edges , respectively ; see fig .
[ fig : model_geom ] ) . for the @xmath31 shift
, @xmath27 is assumed zero for @xmath32 , which means particles are lost when accelerated beyond @xmath33 .
after the @xmath31 shift , at each @xmath10 gridpoint we add to @xmath27 an initial maxwellian with a density chosen to cancel the change in number caused by the @xmath31 shift .
the transverse light - wave variables @xmath34 and @xmath35 advect to the right and left , respectively .
we advance @xmath36 without dispersion ( in vacuum ) by shifts of one @xmath10 gridpoint , which locks the @xmath10 and @xmath37 spacings by @xmath38 .
the extreme edges of the simulation box , beyond the periodicity points @xmath39 of @xmath27 , are `` moat '' regions where @xmath40 for all time , but @xmath41 propagate until leaving the system .
we specify @xmath42 at the boundary from which each field is advected into the box , and can thereby inject a ( pump , seed ) light wave via @xmath43 . due to the low numerical noise in our vlasov code , srbs
does not develop if only the pump light wave is incident and the damping regions are used ( although srbs can occur with no damping regions ) .
we therefore include a low - amplitude srbs seed with a frequency that satisfies linear phase - matching based on the kinetic epw dispersion relation .
the physics is similar in runs where the seed was turned off after kinetically - enhanced srbs develops .
simulations with conditions similar to the trident single hot spot experiments @xcite reveal a sudden onset of kinetically - enhanced srbs as the pump strength increases . a rapid increase in reflectivity
was also observed in the experiments .
we use a pump vacuum wavelength @xmath44 nm , background electron density @xmath45 ( @xmath46 ) where @xmath47 is the pump critical density ( @xmath48 ) , and an electron temperature @xmath49 kev .
the central flattop ( between the krook damping regions ) has length 75.4 @xmath50 m .
we inject an srbs seed with @xmath51 nm ( @xmath52 ) and intensity @xmath53 at the right edge , where @xmath54 is the pump intensity at the left edge ( in this paper , a constant @xmath55 independent of @xmath10 and @xmath37 is understood to be @xmath54 ) .
the pump and seed light waves beat to drive an epw with @xmath56 . in all runs we set @xmath57 and @xmath31 grid spacing @xmath58 .
@xmath59 is the trapping island half - width in the epw electric field , @xmath60 is the trapped electron bounce frequency , and @xmath61 is the epw density perturbation peak amplitude .
@xmath62 is @xmath63 for the epw produced by beating of the pump and unamplified seed light waves ; in other words , @xmath64 resolves trapping even if there were no seed amplification .
the @xmath31 grid extends to @xmath65 for @xmath66 pw/@xmath9 .
[ fig : trid_refavg ] shows @xmath67 versus pump strength , where @xmath67 is the reflectivity @xmath68 time - averaged starting from 1 ps ( to allow the lasers to turn on and transit the domain ) to the end of the run ( at 10 ps ) .
all runs were below the absolute instability threshold @xmath69 pw/@xmath9 for undamped light waves , which corresponds to the threshold undamped srbs growth rate @xmath70 , with @xmath71 the epw landau damping rate .
coupled - mode theory thus predicts srbs approaches a temporal steady state where the seed light wave is convectively amplified across the box .
the steady state can be solved analytically for a three - wave model that includes pump depletion but neglects light - wave damping , and neglects epw advection compared to damping ( the strong damping limit ) @xcite .
the strong damping limit is valid for @xmath72 .
the scattered light wave in this limit satisfies @xmath73 where @xmath74 , @xmath75 , and @xmath76 constant .
@xmath29 is the length of the gain region , and the intensity gain exponent @xmath77 in practical units is @xmath78 { i_{0l,\mathrm{pw}}\lambda_{0,\mu}^2 \over 871}{l\over\lambda_2}.\ ] ] @xmath79 is the scattered light wavelength in the plasma , @xmath80 is @xmath81 in microns , and @xmath82 is @xmath54 in pw/@xmath9 .
@xmath83 $ ] is the electron susceptibility , where @xmath84 is the plasma dispersion function . @xmath85 and @xmath86
are given by the beating of the chosen pump and scattered light waves .
the resulting reflectivity is @xmath87\ ] ] with @xmath88 and @xmath89 .
time - averaged reflectivity @xmath90 and bounce number @xmath91 for trident parameters described in text .
asterisks are simulation @xmath90 .
the solid curve is the coupled - mode steady - state @xmath90 from eq .
( [ eq : tangr ] ) , and the dashed curve is @xmath91 from eq .
( [ eq : nbpd ] ) ; both include pump depletion . ] for weak pumps the simulation gains are small and in accord with the coupled - mode calculation , but they suddenly rise well above it for @xmath92 pw/@xmath9 ( kinetic inflation or enhancement ) .
@xmath90 barely increases with pump strength and saturates around 1% for @xmath93 pw/@xmath9 .
this is well below the coupled - mode level for the largest pump strengths and indicaties nonlinearities besides pump depletion saturate srbs ( kinetic deflation ) .
we do not expect trapping to be important in srbs unless a resonant electron undergoes at least one bounce cycle before crossing the gain region .
this happens for all the runs in fig .
[ fig : trid_refavg ] , using the convective steady - state fields .
the above criterion treats longitudinal endloss as the only de - trapping mechanism , since we have not included transverse sideloss ( e.g. , via the krook operator ) in our runs .
morales and oneil calculate the time - dependent damping rate and frequency of an undriven epw as an initial value problem @xcite .
that is , the damping and frequency vary with the number of bounce orbits the trapped particles have experienced , defined below as @xmath91 .
they find that after the resonant electrons complete about one bounce cycle , landau damping is substantially reduced , and the frequency shift is roughly its late - time asymptotic value .
these results easily translate to a boundary value problem , appropriate to our finite geometry .
however , driven epw s in srbs may differ from morales and oneil s free - wave calculation : see the discussion at the end of this section . in steady state , resonant electrons emerge from a maxwellian distribution at the left damping region ( @xmath94 ) and undergo @xmath95 bounce orbits ( dependent on the field amplitude ) before reaching the position @xmath10 . across a domain of interest
@xmath96 $ ] , @xmath97 where @xmath98 is the bounce wavenumber in the local field @xmath99 .
the total amplitude ( drive plus response ) of the epw driven by two beating light waves as given by the linearized vlasov equation is @xmath100 @xmath101 and @xmath102 is the complex electron oscillation velocity in light wave @xmath103 . a fourier amplitude @xmath104 is related to its physical field @xmath105 by @xmath106 .
in the convective steady state described above , @xmath107 where @xmath108 is @xmath61 for @xmath109 and @xmath110 . the resulting @xmath91 is @xmath111 where @xmath112(n_{2s}/n_0)^{1/2}$ ] is @xmath95 if there were no seed amplification . neglecting pump depletion for illustration ,
@xmath113 $ ] , and @xmath114 , or the total @xmath91 across the domain of length @xmath29 , is @xmath115 note that @xmath116 depends very weakly on the seed intensity .
@xmath114 in the convective steady state is shown in fig .
[ fig : trid_refavg ] as a function of pump strength .
substantial bouncing occurs even for the weakest pump ( @xmath114=1.63 for @xmath55=0.1 pw/@xmath9 ) , suggesting that the convective profiles may be altered by a damping rate and frequency that vary appreciably across the plasma with @xmath95 .
however , kinetic enhancement only occurs for @xmath117 , with @xmath114=3.52 for the smallest enhanced run ( @xmath55=0.4 pw/@xmath9 ) .
this runs counter to the simple - minded expectation , based on the morales - oneil free - wave theory , that srbs inflation should occur for @xmath118 ( or even less , since the damping rate in fig .
1 of @xcite first becomes zero for @xmath119 ) .
this may indicate a difference between the damping rates for free and driven waves .
it could also be that inflation requires the damping to be small over a sufficiently large region , and not just near the right edge .
these questions merit further research .
we now study in detail the case @xmath120
pw/@xmath9 to expose the physics of kinetic enhancement and eas . figure [ fig : trid_reft ] displays the instantaneous reflectivity , with time average @xmath121 that greatly exceeds the coupled - mode level from eq . ( [ eq : tangr ] ) of @xmath122 .
moreover , @xmath123 occurs in temporal sub - picosecond bursts instead of becoming steady , as has also been reported in pic simulations ( e.g. , refs .
@xcite ) .
instantaneous reflectivity @xmath30 for @xmath120 pw/@xmath9 trident run . ]
the time - resolved spectrum of reflected light is displayed in fig .
[ fig : trid_emwall](a ) .
srbs is the dominant signal .
early in time it occurs at the seed frequency @xmath124 but upshifts once srbs becomes strongly enhanced around 3 ps .
this results from the trapping - induced epw frequency downshift @xcite . scattered light between the srbs and pump frequencies is observed shortly after the upshift .
the time - integrated power spectrum , contained in fig . [ fig : trid_emwall](b ) , reveals eas is much stronger than the `` noise level '' in neighboring frequencies ( the noise is partly due to srbs sidelobes resulting from windowing ) .
the eas reflectivity is @xmath125=2.06@xmath126 , from the reflected light in the band @xmath127 during the time eas is observed ( @xmath128 ps ) .
the broad spectrum of reflected light for @xmath129 , larger than the signal above the srbs peak , is scattering off electrostatic fluctuations ( possibly from bams ) with @xmath130 visible in figs .
[ fig : trid_exwk_all](a ) and [ fig : trid_exwk_roots ] .
( color . ) ( a ) time - resolved and ( b ) time - integrated spectrum of reflected light @xmath35 for @xmath120 pw/@xmath9 trident run .
( @xmath131 , @xmath124 ) are the imposed ( pump , seed ) frequency .
the @xmath132 spectrum of a field @xmath104 in `` power db '' is @xmath133 .
the eas asterisk in ( b ) is at @xmath134.,title="fig : " ] + ( color . )
( a ) time - resolved and ( b ) time - integrated spectrum of reflected light @xmath35 for @xmath120 pw/@xmath9 trident run .
( @xmath131 , @xmath124 ) are the imposed ( pump , seed ) frequency .
the @xmath132 spectrum of a field @xmath104 in `` power db '' is @xmath133 .
the eas asterisk in ( b ) is at @xmath134.,title="fig : " ] all power spectra in this paper were made using welch s method of averaged overlapping modified periodograms , with kaiser window functions @xcite .
aggressive windowing was needed to reveal weak , time - dependent signals , such as eas in the presence of much larger srbs .
the low noise inherent in vlasov codes allows such weak processes to be captured .
we now turn to the electrostatic activity .
[ fig : trid_extx ] presents the envelope of the longitudinal electric field , found by rms - averaging @xmath135 over one srbs wavelength and period .
once srbs enters the enhanced regime , the plasma waves propagate away from the laser entrance along fairly well - defined `` rays '' with group velocities @xmath136 .
three broad `` bursts , '' which are strong near the laser entrance around 3 , 6 , and 10 ps , correspond to reflectivity bursts in fig .
[ fig : trid_reft ] .
the initial break - up of the envelope into rays , near the laser entrance shortly after 1 ps , may be due to the trapped particle instability @xcite , and coincides with the first , isolated spike in reflectivity at 1.3 ps .
in particular , the envelope is modulated at @xmath137 , which is close to the bounce wavenumber @xmath138 calculated for the observed amplitude @xmath139 . also , the electrostatic spectrum over the space - time region of the break - up ( not shown ) possesses sidebands at @xmath140 .
however , we do not see clear signs of the trapped particle instability for later times .
( color . )
envelope ( rms ) of electrostatic field @xmath135 for @xmath120 pw/@xmath9 trident run over the @xmath10 domain between the damping regions . ] the electrostatic @xmath7 spectrum , taken over all time and between the two krook damping regions , is shown in fig .
[ fig : trid_exwk_all ] .
the `` stokes '' curve is the locus of daughter electrostatic modes @xmath141 phase - matched for electromagnetic decay of the pump , namely , @xmath142 , with @xmath143 $ ] and @xmath144 for @xmath145 .
@xmath146 and @xmath147 give , respectively , the smaller @xmath85 ( forward srs ) and larger @xmath85 ( backward srs ) legs of the curve .
srbs occurs along the stokes curve , mostly on a frequency - downshifted streak relative to the matched epw .
while there is some activity on the epw curve for @xmath148 , the curve splits into upper and lower branches near @xmath149 .
the extended activity lower in frequency than the epw curve , and which becomes acoustic ( @xmath6 ) with a larger slope for small @xmath150 , is analogous to the beam acoustic modes ( bams ) found by l. yin _ et al .
_ @xcite .
a low - amplitude acoustic mode ( @xmath6 ) with phase velocity @xmath136 , which we call the electron acoustic wave ( eaw ) , is also present .
it is mostly energized for @xmath151 , and remains acoustic but weakens toward higher @xmath150 .
the most strongly - excited eaw has a frequency @xmath152 well above the ion plasma and ion acoustic wave frequencies , so eaw interaction with ions can be neglected .
spectra over successive time windows reveal the eaw is first excited at , and is always strong for , lower @xmath150 .
the eaw and eas both develop after kinetic enhancement of srbs begins , which suggests they result from srbs - induced modifications of the electron distribution .
the asterisk at @xmath153 with phase velocity 1.02@xmath154 is phase - matched to the eas asterisk at @xmath134 in fig .
[ fig : trid_emwall](b ) .
however , the eaw curve is not generated by eas , since most of the eaw energy is far from the phase - matched point .
also , the amplitude is not increased near the phase - matched point , implying decay of the pump laser does not significantly change the eaw level .
the scattering from the eaw appears to be off independently - generated fluctuations and can be viewed as electron acoustic thomson scatter ( eats ) in analogy with thomson scattering from conventional plasma waves .
we propose that a three - wave interaction involving different parts of the extended bam feature generates the low-@xmath150 eaws , and dub this process beam acoustic decay ( bad ) .
the rest of the eaw curve is weakly excited by eaw harmonic coupling ( harmonics of an acoustic mode lie on the same acoustic curve ) .
bad is likely a two - pump process ( instead of a pure parametric decay ) , where the bam curve is energized separately from bad , and bams with different @xmath7 then beat to produce eaws .
this is supported by the fact that the low-@xmath150 bam tail does not have a noticeable peak corresponding to the daughter bam in bad .
moreover , one can estimate the ratio of daughter - wave amplitudes expected from coupled - mode theory . for
weakly damped bams and a strongly - damped eaw ( appropriate for the linear modes found in sec .
[ subsec : modes ] ) , and neglecting pump depletion , the ratio of daughter bam to daughter eaw actions is @xmath155 where @xmath156 is the eaw damping rate and @xmath157 is the bad growth rate . for undamped daughters the actions are equal ( the traditional manley - rowe relation ) .
the @xmath158 power ratio from fig . [ fig : trid_exwk_all ] is very large : @xmath159 .
still , without formulae for @xmath157 and how electric fields relate to actions , it is difficult to infer that little of the bam is produced by bad .
we discuss the generation of bams below .
the eaw and resulting eas occur for moderate pump strengths .
neither appear when srbs is not kinetically enhanced .
both increase with @xmath55 above the enhancement threshold , until no distinct eas peak is visible for @xmath160 pw/@xmath9 .
the upshift in @xmath161 due to the nonlinear downshift of the srbs epw grows significnatly with pump strength . as the pump increases ,
there is also more reflected light higher in freqnuency than srbs ; however , this light is spectrally broad and has no peaks ( such as eas ) .
a distinct eaw is seen for the @xmath162 pw/@xmath9 run but not for @xmath163 pw/@xmath9 .
we discuss the relation of our results to experimental reports of seas in the conlcusion .
we now examine the electron distribution @xmath164 and the linear electrostatic modes it supports .
phase - space vortices develop in the electric field of the srbs plasmon and flatten the space - averaged @xmath164 near @xmath4 . although the vortices are regular for early times , once srbs becomes enhanced they smear into each other .
this indicates a non - monochromatic spectrum and has been seen to result , e.g. , from the trapped particle instability in refs .
figure [ fig : trid_ftplog](a ) displays @xmath164 space- and time - averaged over periods of 6.3@xmath165 and from 50.6 - 53.8 @xmath50 m ( near the domain center ) for the @xmath120 pw/@xmath9 trident run . the plateau width is somewhat correlated to the local wave amplitude , but @xmath164 remains flattened even when the field is weak .
this suggests nonlocality may be important in kinetically - enhanced srbs .
the averaged @xmath164 , taken from fig .
[ fig : trid_ftplog](a ) at 3.15 ps , is plotted in fig .
[ fig : trid_ftplog](b ) and is fairly representative of @xmath164 during enhanced srbs .
the observed flattened plateau half - width ( for the time and space region in fig .
[ fig : trid_ftplog](b ) ) of 1.5@xmath154 compares well with @xmath166 from the observed amplitude @xmath167 .
note that @xmath168 in the trapping region is nearly flat ( it actually increases by 1.6% over its minimum for @xmath169 ) .
this is typical for fully - enhanced srbs , although in the first raman burst there is a clear period when @xmath170 for @xmath169 .
such an upturn appears as the pinched bulge in fig . [
fig : trid_ftplog ] near @xmath171=(2 ps , 4@xmath154 ) .
some adiabatic calculations of trapping in epws show that @xmath164 should be symmetric about the phase velocity when the wave amplitude decreases and electrons are detrapped @xcite .
( color online . ) @xmath164 space- and time - averaged over @xmath172 and @xmath173 50.6 - 53.8 @xmath50 m for @xmath120 pw/@xmath9 trident run .
panels ( a ) and ( b ) are , respectively , for all times and @xmath174 0 , 3.15 ps .
@xmath175 is the phase velocity of the epw driven by the beating of the pump and seed light waves .
@xmath176.,title="fig : " ] + ( color online . )
@xmath164 space- and time - averaged over @xmath172 and @xmath173 50.6 - 53.8 @xmath50 m for @xmath120 pw/@xmath9 trident run .
panels ( a ) and ( b ) are , respectively , for all times and @xmath174 0 , 3.15 ps .
@xmath175 is the phase velocity of the epw driven by the beating of the pump and seed light waves .
@xmath176.,title="fig : " ] figure [ fig : trid_df ] shows @xmath177 at @xmath178 ps , where @xmath179 is a maxwellian fit to @xmath164 for @xmath180v_{te}$ ] ( close to the initial maxwellian ) .
note @xmath181 consists of a broad , beam - like structure near @xmath4 due to trapping , as well as at two lower velocities . modeling @xmath164 as a bulk - and - beam distribution , such as a bi - maxwellian , may not be correct , especially for modes like eaws with low phase velocities near the secondary structures .
@xmath182 ( solid ) and its gauss - hermite projection ( dotted ) for @xmath183 from fig .
[ fig : trid_ftplog ] at @xmath178 ps .
projection parameters given in text . ] to gain insight into the electrostatic dynamics , we study the linear modes of the numerically - obtained @xmath164 .
although a large - amplitude electric field is present and may demand a nonlinear calculation , linear theory yields good agreement with the observed spectrum .
we project @xmath181 onto a gauss - hermite basis , as described in the appendix .
this gives an analytic longitudinal susceptibility @xmath184 , valid in the complex plane , as the sum of @xmath184 for @xmath179 and the projected @xmath181 .
figure [ fig : trid_df ] displays the full @xmath181 and its projection up to the @xmath185 basis function and with the scaled , shifted velocity @xmath186 . the projection only misses fine - scale features which would appear in higher - order basis functions .
these features should weakly distort the real frequency , which stems from collective oscillations in @xmath164 , but may affect the damping rate more strongly , which depends on the slope of @xmath164 at the phase velocity .
the electrostatic linear modes are obtained by solving the dispersion relation @xmath187 for complex @xmath132 given real @xmath150 .
we do not include a ponderomotive term due to parametric coupling with light waves , which would allow for srbs .
[ fig : trid_chi ] presents the zero contours of @xmath188 and @xmath189 , which intersect at the linear modes @xmath190 , for @xmath164 from fig .
[ fig : trid_df ] .
this plot over the same @xmath7 region only using @xmath179 has two zero contours , which intersect at the linear epw ; there is an infinity of strongly - damped modes for the maxwellian , which fall below the chosen plot range . as @xmath191
is increased the contours change slightly , and a single pair of @xmath192 and @xmath193 contours can split into two pairs .
however , the plot remains qualitatively the same .
this is further discussed in the appendix .
zero contours of the electrostatic dispersion relation @xmath190 for @xmath194 , computed from @xmath164 as projected in fig . [ fig : trid_df ] . ]
figure [ fig : trid_froots ] graphs five modes , calculated for @xmath164 from fig .
[ fig : trid_df ] ( @xmath178 ps ) , that resemble the observed spectrum . in fig .
[ fig : trid_exwk_roots ] , @xmath195 for these modes is superimposed on @xmath196 computed for @xmath197 ps ; the agreement is quite good .
the linear modes show the splitting of the epw curve into an upper branch and a set of bams , as well as the appearance of an eaw .
srbs occurs along the mode labeled bam1 , which is frequency - downshifted and less damped than the maxwellian epw .
thus , much of the nonlinear epw physics of morales and oneil is recovered by linear theory with the modified @xmath164 , as also seen in @xcite .
electrostatic linear modes from the projection in fig .
[ fig : trid_df ] .
`` epw '' indicates the epw for the initial maxwellian . ]
( color . ) roots from fig .
[ fig : trid_froots ] overlaid on @xmath196 for @xmath197 ps and the @xmath10 domain of fig .
[ fig : trid_exwk_all ] .
the white and blue circles are possible phase - matched daughter bams and eaws , respectively , for beam acoustic decay of a parent bam1 root with @xmath198 . ] the upper branch and bams are linearly unstable for some @xmath150 , _ in the absence of coupling to light waves .
_ since the slope of the modified @xmath164 is small or slightly positive for some @xmath31 , it is reasonable to find bump - on - tail or beam - plasma instabilities .
these beam - driven , growing electrostatic modes have been studied previously @xcite , and may account for the higher bam fluctuations compared to the level of @xmath148 conventional epws in fig .
[ fig : trid_exwk_all ] .
the linear instability of bams allows them to be excited separately from bad , which based on the discussion at the end of sec .
[ sec : trides ] seems to have little effect on the bams .
the linear modes also contain an eaw with @xmath199 decreasing from @xmath200 to @xmath154 as @xmath150 increases .
it is always heavily damped , unlike the undamped nonlinear eaw due to trapping discussed by other workers ( e.g. , refs .
an eaw also appears in an analogous mode calculation for a bi - maxwellian @xmath164 presented in ref .
@xcite , with a beam whose drift velocity is close to that of the eaw .
it is shown there that stimulated ( quasi - mode ) scattering off the eaw has a small , positive growth rate .
possible phase - matched bad triplets are shown in fig .
[ fig : trid_exwk_roots ] .
note that the lowest-@xmath150 daughter bams ( white circles ) lie above the `` knee '' in the bam feature , and match with daughter eaws ( blue circles ) higher in @xmath150 than the strongest eaw activity .
the break in the bam streak may thus explain the cutoff in the eaw spectrum , although a calculation of the nonlinear coupling mechanism ( e.g. , coupling coefficients ) would shed more light .
to further support the three - wave processes of bad and eats , we turn to bispectral analysis @xcite .
this has been used previously to study three - wave interactions in plasma physics , mostly in magnetized plasmas ( ref .
@xcite is an early example ) .
although we do not pursue it here , bispectral techniques can give estimates of the coupling coefficients and mode growth rates @xcite .
the power spectrum , or the fourier transform of the two - point correlation function , can be generalized to correlations among more signals at more times , or higher - order spectra . given three real , zero - mean signals
@xmath201 , @xmath202 , and @xmath203 , their correlation function @xmath204 and complex bispectrum @xmath205 are @xmath206 and @xmath207 the notation @xmath208 indicates the fields being correlated , and eq .
( [ eq : p3j ] ) shows why @xmath10 is treated differently .
we compute the bispectrum by a generalization of welch s method for power spectra @xcite .
namely , we divide time into @xmath191 windows , apply a window function to the signals , find their fourier transforms , and in each window form @xmath209 @xmath210 where @xmath211 denotes averaging over the @xmath191 windows . @xmath205
measures the signal amplitudes that frequency match ( as three - wave interactions do ) .
if the signals at @xmath212 , @xmath161 , and @xmath86 have independent phases @xmath213 but constant amplitudes , then the @xmath214 s have random phase factors and @xmath215 .
this corresponds to three independent signals that `` gratuitously '' frequency match , although they are not dynamically coupled .
however , for a perfectly coherent process like a pure three - wave interaction , @xmath216 , the @xmath214 s are in phase , and @xmath217 .
the degree of coherence is quantified by the bicoherence @xmath218 : @xmath219 @xmath220 always lies between zero ( no phase coherence ) and one ( perfect phase coherence ) , and its denominator scales like @xmath191 .
@xmath221 is shown in fig .
[ fig : bispec_srs_2d ] .
the diagonal streak indicates stokes decay of the pump : @xmath222 .
srbs is clearly dominant , and weaker eats and scattering off bam noise for @xmath223 are also present .
figure [ fig : bispec_srs_1d ] displays @xmath205 and @xmath218 along the line @xmath224 ; eats is seen in both as a pair of peaks below 6@xmath225 . given the incoherent , thomson - like nature of eats ,
it is not surprising that it is much less bicoherent than srbs .
( color online . )
bispectrum @xmath226 for @xmath227 ps at @xmath228 m for @xmath120 pw/@xmath9 trident run
. the asterisk marks the seeded linear srbs point .
@xmath229 along the dominant diagonal streak . ]
( a ) bispectrum @xmath230 and ( b ) bicoherence @xmath220 from fig .
[ fig : bispec_srs_2d ] along the line @xmath224 for decay of the pump laser . ]
to examine bad , we consider the bispectrum of three spatial fourier amplitudes @xmath231 , computed from @xmath135 . to distinguish left- from right - moving waves , we perform a 2d fourier transform to the @xmath7 domain , set the result to zero for all @xmath232 , and then invert the temporal transform .
this procedure can be considered as a fourier transform in @xmath10 and a hilbert transform in @xmath37 . for a wave @xmath233
this gives @xmath234 .
we use @xmath235 to compute @xmath205 , which requires real fields .
figure [ fig : bispec_bad ] presents @xmath236 and the corresponding @xmath218 , where @xmath237 .
the time window of 5 - 10 ps is when eaw activity is strongest , although similar results obtain for 0 - 5 ps using slightly different @xmath150 s .
@xmath238 is chosen slightly below the maximum eaw power , and @xmath239 gives a larger @xmath205 and more well - defined peak in @xmath218 than do @xmath240 s closer to the srbs peak of @xmath241 .
the strongly localized peak in @xmath205 indicates beam acoustic decay , and persists ( although at lower absolute amplitude ) for nearby @xmath150 choices .
a bicoherence peak , however , is seen only for certain @xmath150 s , which shows when the eaw is driven by the beating of the bams as opposed to satisfying the sum rule with less dynamical coupling .
the frequencies @xmath242 where @xmath205 peaks give @xmath7 pairs that lie , respectively , slightly below the srbs peak , near the `` knee '' in the bam feature , and near the maximum eaw amplitude .
fourier modes @xmath243 0.344 , 0.209 , and 0.136 for @xmath244 , over @xmath227 ps in the @xmath120 pw/@xmath9 trident run . @xmath245 .
the peak in @xmath230 and @xmath220 shows parametric coupling in the bad process . ]
to demonstrate that kinetically - enhanced srbs , bams , eaws , and eats are not peculiar to a narrow parameter range , we show in this section that they occur in ignition hohlraum fill plasmas as well as the single hot spot conditions studied above .
we choose a pump wavelength @xmath246 nm , background density @xmath247 , electron temperature @xmath248 kev , and central krook - free flattop length of 75.2 @xmath50 m .
the srbs seed light wave has @xmath249 nm and @xmath250 , which beats with the pump to produce a plasmon with @xmath251 ( similar to the above trident value ) .
the time - averaged reflectivity @xmath90 versus @xmath55 is plotted in fig .
[ fig : hohl_refavg ] and shows an enhancement threshold of @xmath252 pw/@xmath9 , corresponding to @xmath253 bounce orbits in the convective steady state . note that here @xmath90 for large @xmath55 saturates near 17% , which is much larger than the trident value of 1% .
but for the hohlraum conditions described in the text . ]
we focus on the case @xmath120 pw/@xmath9 .
figure [ fig : hohl_emwt_exwk](a ) displays the time - resolved reflected light spectrum . as for
the trident runs , the srbs light upshifts in frequency , at which time eas develops .
the srbs reflectivity is temporally bursty with large pulses separated by about 1.5 ps . the electrostatic envelope ( not shown ) is similar to fig . [
fig : trid_extx ] , with four broad pulses propagating away from the laser entrance and correlated to reflectivity peaks .
the electrostatic spectrum @xmath196 is presented in fig . [ fig : hohl_emwt_exwk](b ) .
again , the epw curve splits into an upper branch and a set of bams , and srbs is downshifted in @xmath132 along the stokes curve .
an eaw is present and mostly energized at @xmath132 much lower than the eas stokes point .
( color . ) ( a ) time - resolved spectrum of reflected light @xmath35 for @xmath120 pw/@xmath9 hohlraum run .
@xmath254 labels the peak of the eas spectrum , and `` re - scat '' indicates raman re - scatter of srbs in its own forward direction .
( b ) power spectrum of @xmath158 over the @xmath255 domain inside the damping regions .
the stars are stokes points for eas with scattered light of @xmath256 = 2.47 and 2.69 . ] as the pump strength @xmath55 is raised , a distinct eaw and eas persist .
this is unlike the trident runs , where both are washed out by very broad epw frequency downshifts for strong - pump cases .
in addition , reflected light more intense than eas appears near @xmath131 for the hohlraum parameters ( visible in fig .
[ fig : hohl_emwt_exwk](a ) ) , and becomes stronger with increasing @xmath55 .
this can not be stimulated brillouin scattering since the ion are fixed , although mobile ions may significantly distort this signal .
the stokes curve passes through a broad , low - frequency feature near @xmath257 in the @xmath158 spectrum , and the @xmath258 reflected light may be scattering off this .
kinetic inflation of srbs and related physics has been studied with eulerian vlasov simulations in single - hot - spot and hohlraum - fill conditions .
inflation of the reflectivity above the coupled - mode steady state level develops suddenly as the pump becomes stronger , in accord with recent trident experiments .
the trapped particle instability may modulate and break up the electrostatic field envelope early in time but does not dominate the subsequent dynamics . once trapping flattens the distribution , enhanced srbs occurs in picosecond - scale bursts accompanied by electrostatic pulses propagating away from the laser entrance .
srbs light upshifts in frequency due to trapping .
this can be viewed as due to a nonlinear downshift of the daughter plasmon frequency , following morales and oneil , or as scattering off a lower - frequency linear mode supported by the trapping - modified @xmath164 ( in the spirit of quasilinear theory ) .
we adopt the latter approach , and find linear bams that agree with the observed electrostatic spectrum and give frequency - downshifted srbs plasma waves .
our gauss - hermite linear analysis reveals a set of bams , some of which are unstable without light - wave coupling and thus excitable by purely electrostatic dynamics .
eaws and eas have been observed in our simulations and understood in new ways .
the gauss - hermite method reveals the modified @xmath164 supports a heavily - damped _ linear _ eaw , which differs from the nonlinear , undamped eaw reported by rose and others .
it is premature to say that eaws generated by srbs are a solely linear phenomenon or that the nonlinear theories are irrelevant .
linear explanations nonetheless have a certain appeal ( occam s razor ) .
future work should explore the impact of de - trapping mechanisms like speckle sideloss , and broadband seeding in vlasov simulations , on eaw physics .
preliminary results of simulations we have done with a broadband seed show the eas level is independent of bandwidth .
this supports our eats interpretation of eas , as opposed to a stimulated process where seed light in that frequency range would be parametrically amplified .
the connection between our eas picture and experiment is not yet clear .
the relevant trident experiments ( ref .
@xcite , fig .
10 ) for @xmath162 pw/@xmath9 saw eas light of @xmath259 the srbs intensity , similar to the ratio in our run with @xmath120 pw/@xmath9 . for pumps above 5 pw/@xmath9
the epxeriments see spectrally - narrow and much stronger eas ( @xmath260 the srbs intensity ) , while our simulations for @xmath261 pw/@xmath9 show no eas peak but instead strongly - upshifted srbs and broad , higher - frequency light ( as mentioned at the end of sec .
[ sec : trides ] ) .
moreover , earlier measurements on trident@xcite with a random phase plate ( rpp ) smoothing the interaction beam ( giving multiple speckles ) , which were interpreted in ref .
@xcite as potentially being seas , are much stronger than in our runs .
simulations with different density and temperature ( and thus @xmath2 ) from this paper , but still near the trident single - hot - spot values , develop eaws and eas qualitatively like the results presented here ( in particular , the eas intensity is @xmath262 that of srbs ) .
physics not included in this paper s simulations , such as sideloss , ions , or multi - dimensional effects , may provide better experimental contact .
measuring the electrostatic spectrum in the eaw region of the @xmath7 plane would shed light on eaws and eas , in particular revealing if most eaw activity is at lower @xmath150 and @xmath132 than the eas matching point ( as in our runs ) .
our picture of eaw excitation is beam acoustic decay ( bad ) , or the three - wave interaction of two bams and an eaw .
given the very low eaw amplitudes , this is likely a two - pump process , with the daughter bam primarily generated separately from bad .
the daughter eaws can weakly couple to higher-@xmath150 harmonics on the eaw curve .
moreover , @xmath164 is sometimes distorted ( but not fully flattened ) near the eaw phase velocity ; this may lower its damping rate , or facilitate excitation of higher-@xmath150 eaws . the pump laser scatters off the higher-@xmath150 eaw fluctuations , which we think of as electron acoustic thomson scatter ( eats ) .
we emphasize that the eaws are strongest well below the matching point for eas , implying eas can not excite them .
in addition , the eaw amplitude shows no increase at the eas matching point , which again suggests that eas is not significantly energizing the eaws .
after several decades of work , there are many unanswered questions about srbs .
simple estimates for the onset and time - averaged reflectivity due to kinetic inflation are being investigated but are not yet established , and the relative importance of the various saturation mechanisms is still unclear .
research on eaws and eas , besides its intrinsic physical interest , has value in understanding srbs : light scattered off eaws may indicate when distribution functions are strongly modified .
although large eas reflectivity has yet to be clearly observed , it would not be the first surprise in the rich physics of laser - plasma interactions .
we thank w. m. nevins and c. holland for useful discussions of bispectral analysis , and m. m. shoucri for assistance with our vlasov code .
the work at llnl was performed under the auspices of the u. s. department of energy by university of california , lawrence livermore national laboratory under contract no . w-7405-eng-48 .
part of this work was submitted by d. j. s. in partial fulfillment of the requirements for the degree of ph .
d. @xcite and was supported in part by dept . of energy
grant de - fg02 - 91er54109 .
work by a. b. was supported in part by the research lab for electronics and plasma science and fusion center , mit .
the appendix presents how to find the susceptibility and linear modes for an arbitrary distribution by gauss - hermite projection .
we specialize to the 1-d electrostatic modes of an electron plasma with fixed ions , and choose units where @xmath225=@xmath263=@xmath264 for a reference @xmath265 and @xmath266 .
the linear susceptibility @xmath184 is @xmath267 where @xmath268 is the wave phase velocity , the integral is taken along the landau contour , and @xmath104 is the background distribution with @xmath269 . if we write @xmath270 then @xmath271 where @xmath272 is @xmath184 for @xmath273 .
this is convenient for our present work , where @xmath104 consists of a maxwellian plus a small correction centered at a nonzero @xmath274 .
we use a scaled , shifted velocity @xmath275 ( and similarly for @xmath276 ) and put @xmath277 $ ] .
we expand @xmath278 over the basis @xmath279 and obtain @xmath280 where @xmath281 and @xmath282 a convenient basis , which gives an inherently localized @xmath283 and analytically known @xmath284 , is the gauss - hermite basis @xmath285 @xmath286 is the order @xmath287 hermite polynomial ( we follow the notation of @xcite ) . the @xmath288 s are the orthonormal quantum harmonic oscillator eigenstates : @xmath289 .
the projection weights @xmath290 are given by @xmath291 .
the @xmath284 s satisfy a recurrence relation for @xmath292 , found by utilizing properties of @xmath286 : @xmath293 the base cases @xmath294 are @xmath295 @xmath296 , @xmath297 , and @xmath84 is the plasma dispersion function @xcite : @xmath298 we evaluate @xmath84 with the numerical algorithm in ref .
@xcite , which provides a high - order rational function approximation valid throughout the complex plane .
the derivatives of @xmath84 are a polynomial times @xmath84 plus a remainder polynomial .
therefore , we have @xmath299 where @xmath300 and @xmath301 are order @xmath287 polynomials .
( [ eq : chirec ] ) yields a recurrence relation among the polynomials : @xmath302 these formulas hold for @xmath292 and allow the polynomial coefficients to be pre - computed .
the base cases are @xmath303 , @xmath304 , @xmath305 , and @xmath306 .
the linear modes are the roots of the dispersion relation @xmath307 .
we find complex @xmath132 for real @xmath150 using newton s method .
this requires @xmath308 , which from eq .
( [ eq : chipoly ] ) for @xmath292 is @xmath309 @xmath310 and @xmath311 are order @xmath287 polynomials given by ( again for @xmath292 ) @xmath312 the gauss - hermite method has several advantages over other approaches . numerically performing
the landau integral for every @xmath313 of interest is time - consuming , requires care in handling the landau contour , and needs @xmath104 to be analytically continued off the real @xmath313 axis .
our technique only computes several integrals for the @xmath290 s , automatically handles the landau contour via the @xmath84 function , and is valid throughout the complex plane .
writing @xmath104 as a series of maxwellians involves some guesswork since maxwellians do not form an orthogonal basis .
one can use a different scale for @xmath286 and the exponential , for instance @xmath314 .
this gives a simpler recurrence relation for @xmath284 , but requires more terms to reconstruct @xmath104 well than the @xmath288 basis does .
we have tried both bases and find similar linear modes .
unfortunately , the modes given by any gauss - hermite basis do not converge as @xmath315 .
for @xmath316 far inside the classical turning points ( in analogy with the quantum oscillator ) , @xmath317 where @xmath318 , as a wkb analysis easily shows .
higher - order @xmath288 s thus involve oscillation on smaller velocity scales . when extended to complex @xmath316 , these oscillations give rise to exponential growth .
the @xmath284 similarly blow up in the lower half - plane ( as is well - known for the @xmath84 function ) , with @xmath319 growing with @xmath287 for fixed @xmath316 below the real axis .
very small - scale features in @xmath104 project onto high-@xmath287 basis functions @xmath288 , which produce rapidly growing @xmath284 in the complex plane and unphysically distort the modes .
as @xmath191 increases for the parameters of fig .
[ fig : trid_chi ] , the roots vary slightly and a single zero - contour of @xmath320 can bifurcate into two .
low - pass filtering before projection could mitigate this .
although the gauss - hermite method gives much insight into the linear dynamics of an arbitrary @xmath104 , its use requires caution and practice . | 1-d eulerian vlasov - maxwell simulations are presented which show kinetic enhancement of stimulated raman backscatter ( srbs ) due to electron trapping in regimes of heavy linear landau damping .
the conventional raman langmuir wave is transformed into a set of beam acoustic modes [ l. yin _ et al .
_ , phys .
rev .
e * 73 * , 025401 ( 2006 ) ] .
for the first time , a low phase velocity electron acoustic wave ( eaw ) is seen developing from the self - consistent raman physics .
backscatter of the pump laser off the eaw fluctuations is reported and referred to as electron acoustic thomson scatter .
this light is similar in wavelength to , although much lower in amplitude than , the reflected light between the pump and srbs wavelengths observed in single hot spot experiments , and previously interpreted as stimulated electron acoustic scatter [ d. s. montgomery _ et al .
_ , phys .
rev .
lett . * 87 * , 155001 ( 2001 ) ] .
the eaw is strongest well below the phase - matched frequency for electron acoustic scatter , and therefore the eaw is not produced by it .
the beating of different beam acoustic modes is proposed as the eaw excitation mechanism , and is called beam acoustic decay . supporting evidence for this process , including bispectral analysis ,
is presented .
the linear electrostatic modes , found by projecting the numerical distribution function onto a gauss - hermite basis , include beam acoustic modes ( some of which are unstable even without parametric coupling to light waves ) and a strongly - damped eaw similar to the observed one .
this linear eaw results from non - maxwellian features in the electron distribution , rather than nonlinearity due to electron trapping . |
in the past thirty years piezoelectric materials has been widely used as sensors or actuators for the active vibration control of various smart structures like beams @xcite , plates @xcite , and shells @xcite . in typical situations one or two
piezoelectric patches are attached to one or both sides of the elastic structure to be controlled @xcite .
if these patches are excited by an oscillating voltage , they contract or elongate and thereby exert , depending on the configuration of patches and electrodes , membrane force or bending moment on the structure . in the area of bonded patches
the structure changes from homogeneous through the thickness to a sandwich structure with two or three layers : one elastic layer and one or two piezoelectric patches perfectly bonded to it .
modern piezoelectric devices , such as harvesters @xcite , require more complicated piezo patch arrangements .
thus , shape and topology optimization of piezo patches integrated into elastic structures becomes a challenging open field of research , in particular for thin smart structures ( see , eg . , @xcite and the references therein ) . due to the above
mentioned complicated laminate structure , the problems of equilibrium and vibration of smart sandwich plates and shells admit exact analytical solutions of the three - dimensional theory of piezoelectricity only in a few exceptional cases ( see , e.g. , @xcite ) . by this reason
different approaches have been developed depending on the type of the structures .
if smart sandwich plates and shells are thick , no accurate two - dimensional theory can be constructed , so only the numerical methods applied to three - dimensional theory of piezoelectricity make sense @xcite . however , if smart plates and shells are thin , the reduction from the three- to two - dimensional theory is possible and different approximations can be constructed . up to now
two main approaches have been developed : ( i ) the variational approach based on hamilton s variational principle and on some ad - hoc assumptions generalizing kirchhoff - love s hypothesis to smart sandwich plates and shells @xcite , ( ii ) the asymptotic approach based on the analysis of the three - dimensional equations of piezoelectricity , mainly for the plates @xcite .
the disadvantage of the variational approach is the necessity of having an ansatz for the displacements and electric field that is difficult to be justified , while simplicity and brevity are its advantages .
the asymptotic method needs no a priori assumptions ; however , the direct asymptotic analysis of the 3-d differential equations of piezoelectricity is very cumbersome .
the synthesis of these two approaches , called the variational - asymptotic method , first proposed by @xcite and developed further in @xcite , seems to avoid the disadvantages of both approaches described above and proved to be quite effective in constructing approximate equations for thin - walled structures .
note that this method has been applied , among others , to derive the 2-d theory of homogeneous piezoelectric shells in @xcite and the 2-d theory of purely elastic sandwich plates and shells in @xcite .
note also the closely related method of gamma convergence used in homogenization @xcite and plate theories @xcite .
the aim of this paper is to construct the rigorous first order approximate two - dimensional smart sandwich shell theory by the variational - asymptotic method .
we consider the sandwich shell with one elastic layer in the middle and two piezoceramic patches symmetrically bonded to it .
the dimension reduction is based on the asymptotic analysis of the action functional containing small parameters that enables one to find the distribution of the displacements and electric field from the solution of the so - called thickness problem . using the generalized prager - synge identity for the inhomogeneous piezoelectric body ,
we provide also the error estimation of the constructed theory in the energetic norm .
we apply this theory to the problem of forced vibration of a circular plate partially covered by two circular rings of piezoceramic layers with thickness polarization excited by a harmonic voltage and find the exact analytical solution to this problem .
the paper is organized as follows .
after this short introduction the variational formulation of the problem is given in section 2 .
sections 3 and 4 are devoted to the asymptotic analysis of the action functional . in section 5 the two - dimensional theory of piezoelectric sandwich shells is obtained . in section 6
we provide the error estimation of the constructed theory .
section 7 presents the exact analytical solution to the forced vibration of a circular sandwich plate .
finally , section 8 concludes the paper .
let @xmath0 be a two - dimensional smooth surface bounded by a smooth closed curve @xmath1 .
at each point of the surface @xmath0 ( called the middle surface of the shell ) a segment of length @xmath2 in the direction perpendicular to the surface is drawn so that its centre lies on the surface .
if the length @xmath2 is sufficiently small , the segments do not intersect each other and fill the domain @xmath3 occupied by a sandwich shell in its undeformed state .
we analyze the forced vibration of the sandwich shell consisting of three layers shown schematically in fig .
[ fig:1 ] .
the middle layer is made of an elastic ( dielectric ) transversely isotropic and homogeneous material that does not exhibit piezo - effects .
the upper and lower layers are made of the same transversely isotropic homogeneous piezoceramic material with the thickness polarization perfectly attached to the middle one .
these layers can be used as sensor or actuator for the purpose of active control of shell vibration .
the faces of piezoelectric layers are covered by the electrodes whose thickness is neglected .
thus , the undeformed shell occupies the domain specified by the equation @xmath4 where @xmath5 is the equation of the middle surface @xmath0 , and @xmath6 are the cartesian components of the normal vector @xmath7 to this surface .
we shall use latin indices running from 1 to 3 to refer to the spatial co - ordinates and the greek indices running from 1 to 2 to refer to the surface co - ordinates @xmath8 and @xmath9 . the curvilinear co - ordinates @xmath10 take values in a domain of @xmath11 , while @xmath12 $ ] .
let the upper and lower faces of the piezoceramic layers covered by the electrodes ( at @xmath13 ) be denoted by @xmath14 . on these surfaces the electric potential
is prescribed @xmath15 hamilton s variational principle of piezoelectricity ( see , e.g. , @xcite ) states that the true displacement @xmath16 and electric potential @xmath17 of an inhomogeneous piezoelectric body change in space and time in such a way that the action functional @xmath18=\int_{t_0}^{t_1}\int_{\mathcal{v}}[t(\mathbf{x},\dot{\mathbf{w}})-w(\mathbf{x},\boldsymbol{\varepsilon } , \mathbf{e})]\ , dv \ , dt\ ] ] becomes stationary among all continuously differentiable functions @xmath19 and @xmath20 satisfying the initial and end conditions @xmath21 as well as constraint .
the integrand in the action functional is called lagrangian , while @xmath22 is the volume element and the dot over quantities denotes the partial time derivative . in the lagrangian
@xmath23 describes the kinetic energy density given by can be neglected . ]
@xmath24 with @xmath25 being the mass density .
function @xmath26 , called electric enthalpy density , reads @xmath27 where @xmath28 is the strain tensor @xmath29 while @xmath30 the electric field @xmath31 applying the standard calculus of variation one easily shows that the stationarity condition @xmath32 implies the equations of motion of piezoelectric body ( including the equation of electrostatics ) @xmath33 where the stress tensor @xmath34 and the electric induction field @xmath35 are given by @xmath36 we call @xmath37 the ( fourth - rank ) tensor of elastic stiffnesses , @xmath38 the ( third - rank ) tensor of piezoelectric constants , while @xmath39 the ( second - rank ) tensor of dielectric permittivities . for the elastic ( dielectric ) material @xmath40 , so it is the degenerate case of piezoelectric material . substituting the constitutive equations into and making use of the kinematic equations and , we get the closed system of four governing equations for four unknown functions @xmath19 and @xmath20 . for the asymptotic analysis of the sandwich shell it is convenient to use the curvilinear coordinates @xmath41 introduced above and the co- and contravariant index notation for vectors and tensors , with einstein s summation convention being employed . in this coordinate system the action functional reads @xmath42 \kappa \ , dx\ , da\ , dt,\ ] ] where @xmath43 ( with @xmath44 and @xmath45 being the mean and gaussian curvature of the middle surface , respectively ) and @xmath46 denotes the area element of the middle surface .
the kinetic energy becomes @xmath47 where @xmath48 are the contravariant components of the surface metric tensor , while @xmath49 and @xmath50 are the projections of the displacement vector onto the tangential and normal directions to the middle surface @xmath51 the electric enthalpy density @xmath52 reads @xmath53 for the sandwich shells we have @xmath54 @xmath55 thus , the label @xmath56 in @xmath57 indicates elastic stiffnesses at constant electric field , while the label @xmath58 in @xmath59 denotes dielectric permittivities at constant strain for the piezoelectric material .
the material constants with an over - bar correspond to those of the elastic ( dielectric ) material . the problem is to replace the three - dimensional action functional by an approximate two - dimensional action functional for a thin shell , whose functions depend only on the longitudinal co - ordinates @xmath60 and time @xmath61 .
the possibility of reduction of the three- to the two - dimensional problem is related to the smallness of the ratios between the thickness @xmath2 and the characteristic radius of curvature @xmath62 of the shell middle surface and between @xmath2 and the characteristic scale of change of the electroelastic state in the longitudinal directions @xmath63 @xcite .
we assume that @xmath64 additionally , we assume that @xmath65 where @xmath66 is the characteristic scale of change of the functions @xmath67 and @xmath68 in time ( see @xcite ) and @xmath69 the minimal velocity of plane waves in the piezoelectric materials under consideration .
this means that we consider in this paper only statics or low - frequency vibrations of the inhomogeneous piezoelectric shell . by using the variational - asymptotic method , the two - dimensional action functional
will be constructed below in which terms of the order @xmath70 and @xmath71 are neglected as compared with unity ( the first - order or `` classical '' approximation ) . in order to fix the domain of the transverse co - ordinate in the passage to the limit @xmath72 , we introduce the dimensionless co - ordinate @xmath73.\ ] ] now @xmath2 enters the action functional explicitly through the components of the strain tensor @xmath74 and the electric field @xmath75 @xmath76 here the semicolon preceding greek indices denotes the co - ordinate expression for the covariant derivatives on the surface , the raising or lowering of indices of surface tensors will be done with the surface metrics @xmath48 and @xmath77 , @xmath78 and @xmath79 are the second and third fundamental forms of the surface , vertical bar followed by @xmath80 indicates the partial derivative with respect to @xmath80 and _ not _ with respect to @xmath81 .
we denote by @xmath82 the integral over @xmath80 within the limits @xmath83 $ ] .
before applying the variational - asymptotic procedure let us transform the electric enthalpy density to another form more convenient for the asymptotic analysis @xcite .
we note that among terms of @xmath84 the derivatives @xmath85 and @xmath86 in @xmath87 and @xmath88 as well as @xmath89 are the main ones in the asymptotic sense .
therefore it is convenient to single out the components @xmath87 and @xmath88 as well as @xmath90 in the electric enthalpy density .
we represent the latter as the sum of two quadratic forms @xmath91 and @xmath92 corresponding to longitudinal and transverse electric enthalpy densities , respectively .
these are defined by @xmath93 let us first find the decomposition in the most general case of anisotropy @xcite .
long , but otherwise simple calculations show that @xmath94 where @xmath95 the coefficients @xmath96 , @xmath97 , @xmath98 , @xmath99 , @xmath100 , @xmath101 , @xmath102 , @xmath103 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 and @xmath110 can be regarded as components of surface tensors referred to the basis vectors @xmath111 of the middle surface .
we shall call them `` two - dimensional '' electroelastic moduli .
they are evaluated in terms of the three - dimensional moduli by means of the formulas @xmath112 note that , as these tensors are referred to the basis @xmath113 , their components will depend on @xmath80 through the shifter @xmath114 even for homogeneous layers . however , since we construct the approximate shell theory admitting the error @xmath70 , this weak dependence on @xmath80 of the 2-d moduli of the homogeneous layers can be neglected .
besides , for our sandwich shell possessing the transversal isotropy all 2-d tensors of odd rank vanish @xmath115 the 2-d tensors of even rank are given by @xmath116 @xmath117 where @xmath118 and voigt s notation is used on the right - hand side of these formulas ( see @xcite ) . taking into account the above properties , we present the longitudinal and transverse electric enthalpy densities in the form @xmath119 where @xmath120
we restrict ourselves to the low frequency vibrations of the sandwich shell for which assumption is valid . based on this assumption
we may neglect the kinetic energy density in the variational - asymptotic procedure . at the first step of the variational - asymptotic procedure we keep only the asymptotically principal terms in the transverse electric enthalpy densities having the formal order @xmath121 .
since the obtained functional contains only the derivatives with respect to @xmath80 , we drop the integration over @xmath0 and @xmath61 and reduce the problem to finding extremal of the functional @xmath122d\zeta\ ] ] among @xmath49 , @xmath50 , and @xmath68 satisfying constraint where the coordinates @xmath10 and time @xmath61 are regarded as parameters .
it is easy to show that the extremal of satisfies the equations @xmath123 subjected to the boundary conditions @xmath124 the continuity of displacements @xmath125 , @xmath50 and electric potential @xmath68 , as well as the continuity of @xmath126 , @xmath127 , and @xmath128 when crossing the points @xmath129 and @xmath130 . since the functions @xmath131 , @xmath132 , and @xmath133 are piecewise constant in the interval @xmath134 , @xmath135 , and @xmath136 , equations admit an exact integration in those intervals .
the constants of integrations can be found from the boundary conditions and the continuity conditions . omitting the calculations , we present the final results @xmath137 @xmath138}{\bar{\epsilon}_{33}(2\zeta _ 0 - 1)-2\zeta_0\epsilon^s_{33}(1+k_t^2 ) } & \text{for $ |\zeta |>\zeta _ 0 $ } , \end{cases } \notag\end{aligned}\ ] ] where @xmath139 and @xmath140 are arbitrary functions of @xmath10 and @xmath61 and @xmath141 the thickness coupling factor . at the second step of the variational - asymptotic procedure we fix @xmath142 and @xmath143 and seek the stationary point of the functional in the form @xmath144 where @xmath145 is the correction term . by redefining @xmath142 if required
, we can put the following constraint on the function @xmath146 : @xmath147 according to the above constraint , @xmath142 describes the mean displacements of the shell in the longitudinal directions . keeping in the electric enthalpy densities only asymptotically principal terms containing @xmath148
, we arrive at finding the extremal of the following functional @xmath149 obviously the extremum is zero and is achieved at @xmath150 , where @xmath151 describe the rotation angles of the middle surface . at the third step
we look for the stationary point of the functional in the form @xmath152 where @xmath153 , @xmath154 , and @xmath155 are the correction terms . without restricting generality , we can put the following constraints on the functions @xmath156 and @xmath157 @xmath158 with these constraints being fulfilled , we may interpret @xmath139 and @xmath140 as the mean displacements of the shell .
keeping in the electric enthalpy densities only asymptotically principal terms containing @xmath156 , @xmath157 , and @xmath159 , we arrive at the problem of finding the extremal of the following functional @xmath160 \ , d\zeta , \ ] ] where @xmath161 with @xmath162 describing the measures of extension and bending of the shell middle surface , respectively .
varying functional @xmath163 , we obtain the euler equations @xmath164 ( with @xmath165 and @xmath166 being the lagrange multipliers that can be found later from the constraints ) subjected to the boundary conditions @xmath167 the continuity of @xmath156 , @xmath157 , and @xmath159 , as well as the continuity of @xmath168 , @xmath169 , and @xmath170 when crossing the points @xmath129 and @xmath130 .
equations together with the boundary conditions and the continuity conditions constitute the so - called thickness problem that enables one to find the asymptotic distributions of displacements and electric potential for the piezoelectric sandwich shell
. this problem can be solved in the similar manner as that of and . omitting the long , but otherwise standard integration procedure , we present the final results @xmath171 where @xmath172 . functions @xmath173 , @xmath174 , @xmath175 do not depend on @xmath80 , but depend on @xmath10 and @xmath61 through @xmath176 , @xmath177 and @xmath178 .
the formulas for them read @xmath179 , \\
\alpha ( \frac{1}{8}-\zeta _ 0 ^ 3)+(r-\bar{f})hb^\alpha _ \alpha \zeta _ 0 ^ 2(\frac{1}{2}-\zeta _ 0)-\bar{f}hb^\alpha _ \alpha \frac{\zeta _ 0 ^ 3}{3 } , \\ q_2&=-qhb^\alpha
_ \alpha \frac{1}{2}(\zeta _ 0 ^ 2-\frac{1}{4 } ) , \\ q_3&=(r-\bar{f})a^\alpha _
\alpha \zeta _ 0-\frac{r}{3}hb^\alpha _ \alpha ( \frac{1}{8}-\zeta _ 0 ^ 3)-(r-\bar{f})hb^\alpha _
\alpha \zeta _ 0 ^ 3-\bar{f}hb^\alpha _ \alpha \frac{\zeta _ 0 ^ 3}{3 } \\ & + \langle \frac{1}{\epsilon^p_{33}(\zeta ) } \rangle ^{-1 } [ \frac{\varphi _ 0(t)}{h}-qa^\alpha _ \alpha ( 1 - 2\zeta _ 0)]\frac{k_t^2}{e_{33}(1+k_t^2)}\zeta _ 0 , \\ q_4&=-(r-\bar{f})a^\alpha _ \alpha \zeta _ 0-\frac{r}{3}hb^\alpha _ \alpha ( \frac{1}{8}-\zeta _ 0 ^ 3)-(r-\bar{f})hb^\alpha
_ \alpha \zeta _ 0 ^ 3-\bar{f}hb^\alpha
_ \alpha \frac{\zeta _ 0 ^ 3}{3 } \\ & -\langle \frac{1}{\epsilon^p_{33}(\zeta ) } \rangle ^{-1 } [ \frac{\varphi _ 0(t)}{h}-qa^\alpha _ \alpha ( 1 - 2\zeta _ 0)]\frac{k_t^2}{e_{33}(1+k_t^2)}\zeta _ 0 , \\
q_5&=\frac{\varphi _ 0(t)}{2h}-\frac{1}{2}q a^\alpha _ \alpha + \frac{1}{8}q hb^\alpha _ \alpha + \frac{q_0}{2\epsilon^p_{33 } } , \\
0(t)}{2h}+\frac{1}{2}q a^\alpha _ \alpha + \frac{1}{8}q hb^\alpha _ \alpha -\frac{q_0}{2\epsilon^p_{33 } } .\end{aligned}\ ] ] it is interesting to note that both @xmath180 and @xmath181 are piecewise constant and are given by @xmath182
in accordance with the variational - asymptotic method we take now the displacement field and the electric field represented in , where functions @xmath156 , @xmath157 , and @xmath159 are given by .
we regard @xmath142 and @xmath143 as the unknown functions , with @xmath183 and @xmath184 describing the measures of extension and bending of the shell middle surface , respectively .
we substitute this displacement and electric fields into the action functional .
since we construct the approximate theory admitting the error of order @xmath70 , @xmath185 in may be replaced by 1 .
if we keep only the principal terms containing the unknown functions in the average lagrangian and integrate over the thickness , then the average kinetic energy density becomes @xmath186 where @xmath187 to compute the average electric enthalpy density we use the additive decomposition @xmath188 that leads to @xmath189 as @xmath190 is negligibly small on the fields , we may neglect the last term of @xmath91 in @xmath191 and approximate the strains @xmath192 by @xmath193 thus , for the average longitudinal electric enthalpy density the integration over the thickness yields @xmath194 where @xmath195 , \\ \langle c^n_2(\zeta ) \rangle = c^n_2(1 - 2\zeta _
0)+\bar{c}_2^p 2\zeta _ 0 , \quad \langle \zeta
^2c^n_2(\zeta ) \rangle = \frac{1}{12}[c^n_2(1 - 8\zeta _ 0 ^ 3)+\bar{c}_2^p8\zeta _ 0 ^ 3 ] .
\end{split}\ ] ] note that the cross terms between @xmath183 and @xmath184 do not appear in the average electric enthalpy density thank to the symmetric placement of the piezoceramic layers with respect to the middle elastic layer causing the evenness of functions @xmath196 and @xmath197 .
if these layers are not placed symmetrically or if only one piezoceramic layer is bonded , the cross terms between @xmath183 and @xmath184 will certainly appear leading to the cross effects in tension and bending .
for the average transverse electric enthalpy density we use the fact that @xmath198 while @xmath180 and @xmath181 are piecewise constant as described in .
therefore the integral of @xmath199 over @xmath80 can be taken as the sum of three integrals @xmath200 it follows from that , for the piezoceramic layers , @xmath201 while for the elastic ( dielectric ) layer , @xmath202 thus , @xmath203 ^ 2,\ ] ] because for the dielectric material @xmath204 . combining the average longitudinal and transverse electric enthalpy densities together
, we obtain @xmath205 ^ 2.\end{gathered}\ ] ] we formulate now the variational principle for the smart sandwich shell : the average displacement field @xmath206 of the sandwich shell changes in space and time in such a way that the 2-d average action functional @xmath207=\int_{t_0}^{t_1}\int_{\omega } [ \theta ( \dot{\mathbf{u}})-\phi ( a_{\alpha \beta } , b_{\alpha \beta } ) ] \ , da \ , dt\ ] ] becomes stationary among all continuously differentiable functions @xmath206 satisfying the initial and end conditions @xmath208 the standard calculus of variation shows that the stationarity condition @xmath209 implies the following two - dimensional equations @xmath210 and @xmath211 subjected to the free - edge boundary conditions @xmath212 where @xmath213 and @xmath214 denotes the components of the surface vector normal to the curve @xmath215 . for the clamped or simply supported edge
, natural boundary conditions must be replaced by the corresponding kinematical boundary conditions .
the equations of motion and must be complemented by the constitutive equations @xmath216a^{\alpha \beta } \label{5.14 } \\
m^{\alpha \beta } & = \frac{\partial \phi } { \partial b_{\alpha \beta } } = h^3\langle \zeta ^2c^n_1(\zeta )
\rangle b^\lambda _
\lambda a^{\alpha \beta } + 2h^3\langle \zeta ^2c^n_2(\zeta ) \rangle
b^{\alpha \beta } .\end{aligned}\ ] ] note that , if the elastic layer is not fully covered by the piezoelectric patches , the average mass density and stiffnesses of the 2-d theory will suffer jump across the boundary between the covered and uncovered area . in this case the jump conditions along this boundary can also be obtained ( see section 7 ) .
it is readily seen that the constructed two - dimensional theory of sandwich shells reduces to the classical theories of elastic shells ( with @xmath217 ) and piezoelectric shells ( with @xmath218 ) , respectively ( cf .
@xcite ) . to complete the 2-d theory of piezoelectric sandwich shells
we should also indicate the method of restoring the 3-d electroelastic state by means of the 2-d one .
to do this , the strain tensor @xmath219 and the electric field @xmath30 should be found from .
it can be shown by the asymptotic analysis that the following formulas @xmath220 hold true within the first - order approximation .
using , we find that @xmath221 note that , in contrast to the homogeneous piezoelectric shells , the strain @xmath88 and the electric field @xmath90 in the smart sandwich shells are discontinuous through the thickness .
the stress tensor @xmath222 and the electric induction @xmath35 are then determined by the 3-d constitutive equations . while doing so , it is convenient to use the decomposition for the electric enthalpy density . within the first - order approximation
we find @xmath223 . \notag\end{gathered}\ ] ] note that @xmath222 and @xmath35 are continuous through the thickness .
again , these formulae are accurate up to terms of the orders @xmath70 and @xmath71 of smallness .
in this section we shall prove an identity that generalizes prager - synge s identity found in @xcite to the statics of inhomogeneous piezoelectric bodies . based on this identity an error estimate of the smart sandwich shell theory constructed in the previous section for the special case of statics
will be established .
we consider an inhomogeneous piezoelectric body occupying the three - dimensional domain @xmath3 in its undeformed state that stays in equilibrium under a fixed voltage . concerning the boundary conditions for the mechanical quantities we assume that the boundary @xmath224 is decomposed into two subboundaries @xmath225 and @xmath226 .
on the part @xmath227 the displacements vanish ( clamped boundary ) @xmath228 on the remaining part @xmath226 the traction - free boundary condition is assumed @xmath229 concerning the boundary conditions for the electric potential we assume that the boundary @xmath224 consists of @xmath230 subboundaries @xmath231 and @xmath232 .
the subboundaries @xmath233 are covered by electrodes with negligible thickness . on these electrodes the electric potential
is prescribed @xmath234 on the uncoated portion @xmath235 of the boundary we require that the surface charge vanishes @xmath236 we introduce the linear vector space of electroelastic states that consists of elements of the form @xmath237 , where @xmath222 is the stress field and @xmath30 is the electric field ; both fields are defined in the three - dimensional domain @xmath3 occupied by the piezoelectric body . in this space
we introduce the following energetic norm @xmath238=\int_{\mathcal{v } } g(\boldsymbol{\sigma } , \mathbf{e})\ , dv , \label{6.1}\ ] ] where function @xmath239 is the density of the complementary energy ( or gibbs function ) @xcite . in component form @xmath239
reads @xmath240 where the precise dependence of the electroelastic moduli on @xmath241 is suppressed for short .
since the complementary energy density @xmath239 is positive definite , the definition is meaningful .
we call `` kinematically admissible '' those electroelastic states @xmath242 for which the compatible strain field @xmath243 and the electric induction field @xmath244 exist such that @xmath245 while @xmath246 and @xmath247 are expressed in terms of @xmath243 and @xmath244 by the constitutive equations equivalent to .
we call those electroelastic states @xmath248 `` statically admissible '' , when @xmath249 let @xmath250 be the true electroelastic state that is realized in an inhomogeneous piezoelectric body staying in equilibrium under the given values of the electric potential @xmath251 on the electrodes @xmath252
. then the following identity @xmath253= c_2[\frac{1}{2}(\check{\boldsymbol{\xi}}-\hat{\boldsymbol{\xi } } ) ] \label{6.2}\ ] ] turns out to be valid for arbitrary kinematically and statically admissible fields @xmath242 and @xmath248 .
this identity generalizes the well - known prager - synge identity @xcite to the statics of inhomogeneous piezoelectric bodies .
it implies that @xmath254 may be regarded as an `` approximation '' to the true solution in the energetic norm , if the complementary energy associated with the difference @xmath255 is `` small '' . in this case
we may also consider each of the fields @xmath242 or @xmath248 as an `` approximation '' , in view of the inequalities @xmath256\le c_2[\check{\boldsymbol{\xi } } -\hat{\boldsymbol{\xi } } ] , \\
c_2[\tilde{\boldsymbol{\xi } } -\hat{\boldsymbol{\xi } } ) ] \le c_2[\check{\boldsymbol{\xi } } -\hat{\boldsymbol{\xi } } ] , \end{gathered}\ ] ] which follow easily from . to prove the identity we first rewrite its left - hand side as follows @xmath257&= c_2[\tilde{\boldsymbol{\xi } } -\hat{\boldsymbol{\xi } } -\frac{1}{2}(\check{\boldsymbol{\xi } }
-\hat{\boldsymbol{\xi } } ) ] \notag \\ & = c_2[\tilde{\boldsymbol{\xi } } -\hat{\boldsymbol{\xi } } ] + c_2[\frac{1}{2}(\check{\boldsymbol{\xi } } -\hat{\boldsymbol{\xi } } ) ] -[\tilde{\boldsymbol{\xi } } -\hat{\boldsymbol{\xi } } , \check{\boldsymbol{\xi } } -\hat{\boldsymbol{\xi } } ] \notag \\ & = c_2[\frac{1}{2}(\check{\boldsymbol{\xi } } -\hat{\boldsymbol{\xi } } ) ] + [ \tilde{\boldsymbol{\xi } } -\hat{\boldsymbol{\xi } } , \tilde{\boldsymbol{\xi } } -\check{\boldsymbol{\xi } } ] , \label{6.3}\end{aligned}\ ] ] where @xmath258 $ ] denotes the scalar product of two elements @xmath259=\int_{\mathcal{b}}g(\boldsymbol{\xi } , \boldsymbol{\xi } ^\prime ) \ , dv = \int_{\mathcal{b}}s_{ab}\boldsymbol{\xi } ^a\boldsymbol{\xi } ^{\prime b}\ , dv .
\label{6.4}\ ] ] in @xmath260 and @xmath261 is a symmetric matrix , whose elements are themselves matrices @xmath262 where @xmath263 are elements of the transpose matrix and voigt s abbreviated index notation is used . according to , the identity holds true , when @xmath264=0.\end{aligned}\ ] ] this identity follows from the definitions of @xmath265 , @xmath242 , @xmath248 and the formulae and .
indeed @xmath264&= \int_{\mathcal{v}}[(\tilde{\boldsymbol{\sigma } } -\hat{\boldsymbol{\sigma } } ) \mathbf { : } ( \tilde{\boldsymbol{\varepsilon } } -\check{\boldsymbol{\varepsilon } } ) + ( \tilde{\mathbf{e}}-\hat{\mathbf{e } } ) \cdot ( \tilde{\mathbf{d}}-\check{\mathbf{d}})]\ , dv \\ & = \int_{\mathcal{v}}[(\tilde{\boldsymbol{\sigma } } -\hat{\boldsymbol{\sigma } } ) \mathbf { : } \nabla ( \tilde{\mathbf{w}}-\check{\mathbf{w}})+\nabla ( \tilde{\varphi } -\hat{\varphi } ) \cdot ( \tilde{\mathbf{d}}-\check{\mathbf{d}})]\ , dv , \end{aligned}\ ] ] which is the consequence of the definitions of @xmath265 , @xmath242 and .
integrating this identity by parts and taking the definition of @xmath248 as well as the boundary conditions into account , we see that the right - hand side vanishes .
thus , the identity is proved .
based on the following error estimate can be established .
_ theorem .
_ the electroelastic state determined by the 2-d static theory of piezoelectric sandwich shells differs in the norm @xmath266 from the exact electroelastic state determined by the 3-d theory of piezoelectricity by a quantity of the order @xmath267 as compared with unity . to prove this theorem it is enough to find out the kinematically and statically admissible 3-d fields of electroelastic states that differ from the electroelastic state determined by the 2-d theory by a quantity of the order @xmath267 as compared with unity .
below we shall construct these fields .
_ construction of kinematically admissible field .
_ we specify the kinematically admissible displacement field in the form @xmath268 where @xmath151 , while @xmath269 , @xmath270 , @xmath271 are given by and , respectively . here and below ,
all quantities without the superscripts @xmath272 and @xmath273 refer to the solution of the equilibrium equations of piezoelectric sandwich shells obtained by the constructed two - dimensional theory .
the components of the strain tensor are calculated according to .
assume that the 2-d electroelastic state is characterized by the strain amplitude @xmath274 .
the asymptotic analysis similar to that given in section 4 shows that @xmath275 with @xmath88 from .
we choose the components @xmath276 of the electric induction to be zero , while @xmath277 it is easy to see that @xmath244 satisfies the exact 3-d equation of electrostatics @xmath278 and , due to the property of @xmath185 , @xmath279 note that the constructed field @xmath244 does not satisfy the exact boundary condition @xmath280 , posed at the portion @xmath281 $ ] of the edge .
for simplicity of the proof we further assume that the 3-d boundary conditions at the edge of the shell agree with the inner expansion of the electroelastic state ( the so - called regular boundary conditions ) .
then the electric induction field @xmath282 constructed above is kinematically admissible . for irregular boundary conditions we have to take into account an additional electric induction field that differs substantially from zero only in a thin boundary layer at the shell edge .
since the energy of this boundary layer is of the order @xmath71 compared with that of the inner domain , one can easily generalize the proof of the theorem to this case .
knowing @xmath283 , we find @xmath284 from the constitutive equations equivalent to . because @xmath285 , it is easily seen that @xmath286 .
_ construction of statically admissible field .
_ we write down the exact 3-d equilibrium equations for a shell in the form ( cf .
@xcite ) @xmath287 where @xmath288 note that @xmath289 is unsymmetric . to find the statically admissible stress field @xmath290 satisfying and the traction - free boundary conditions @xmath291 we proceed as follows .
we specify @xmath292 in the form @xmath293 where @xmath294 and @xmath295 are symmetric and independent of @xmath296 .
these are chosen from the conditions @xmath297 where @xmath298 .
the conditions enable one to determine @xmath294 and @xmath295 through @xmath299 and @xmath300 uniquely . moreover
, one can check that @xmath301 solving , with the given @xmath289 , we can find @xmath302 and @xmath303 and then @xmath304 and @xmath305 .
it turns out that are the sufficient conditions for the existence of @xmath302 and @xmath303 . indeed , integrating and @xmath191 multiplied by @xmath296 over @xmath12 $ ]
, we obtain @xmath306 where @xmath307 . from the first and the last equations of it follows that @xmath308 , since @xmath309 according to the 2-d equations of equilibrium . from the second equation of we also obtain @xmath310 . thus ,
if the boundary conditions are satisfied at @xmath311 , then after the integration they will also be satisfied at @xmath312 . not showing the cumbersome solution of , we note only that @xmath313 .
thus , @xmath314 . concerning the statically admissible electric field
@xmath315 we specify its potential by @xmath316 with @xmath317 and @xmath318 from and , respectively .
then @xmath319 note that the statically admissible field @xmath320 constructed above satisfies only the regular boundary conditions at the shell edge , exactly as in the previous case . from this construction
we see that @xmath242 and @xmath248 differ from those found by the 2-d theory by a quantity of the order @xmath70 and @xmath71 as compared with unity .
we have thus established the asymptotic accuracy of the 2-d theory in the energetic norm .
in this section we illustrate the application of the theory to the problem of axisymmetric longitudinal vibration of an elastic circular plate of radius @xmath321 partially covered by the piezoceramic patches with thickness polarization .
suppose that the piezoceramic patches cover only the rings @xmath322 ( defined by @xmath323 ) of the face surfaces , where @xmath324 are the polar co - ordinates .
when these piezoceramic patches are subjected to an oscillating voltage , the electric field occurs leading to the extension or contraction of the patches and forcing the plate to vibrate in the radial direction . for the axisymmetric longitudinal vibration of the plate
the bending measures vanish , while for the measures of extension we have @xmath325 where @xmath326 is the radial displacement .
thus , the action functional , up to an unimportant factor @xmath327 , becomes the sum of the following integrals @xmath328\varrho d\varrho \ , dt \notag \\
+ \int_{t_0}^{t_1}\int_{r_0}^r \rho ^l h\dot{u}_\varrho ^2 \varrho d\varrho \ , dt-\int_{t_0}^{t_1}\int_{r_0}^r h[c_1^l(u_{\varrho , \varrho } + \frac{u_\varrho } { \varrho } ) ^2 + 2c_2^lu_{\varrho , \varrho } ^2 + 2c_2^p\frac{u_\varrho ^2}{\varrho ^2 } \notag \\ + 2e^l\frac{\varphi _ 0(t)}{h}(u_{\varrho , \varrho } + \frac{u_\varrho } { \varrho } ) ] \varrho d\varrho \ ,
\label{7.0}\end{gathered}\ ] ] we use label @xmath329 to indicate the coefficients in this functional for the sandwich plate with three layers that are functions of @xmath330 @xmath331 @xmath332 varying functional , we obtain the euler equation @xmath333 the jump conditions at @xmath334 @xmath335|_{r_0- } \label{7.2 } \\ & = [ ( c_1^l+2c_2^l)u_{\varrho , \varrho } + c_1^lu_\varrho /\varrho + e^l\frac{\varphi _ 0(t)}{h}]|_{r_0 + } , \notag\end{aligned}\ ] ] and the traction - free boundary condition at @xmath336 @xmath337|_{r}=0 . \label{7.3}\end{aligned}\ ] ] the voltage @xmath178 is assumed to depend harmonically on @xmath61 , @xmath338 , so that solutions of - can be sought in the form @xmath339 introducing the dimensionless variable and quantities @xmath340 this system can be transformed to the differential equations @xmath341 where @xmath342 , the jump conditions at @xmath343 @xmath344|_{\bar{r}- } = [ \hat{u}^\prime + \gamma \hat{u}/y + d\frac{\hat{\varphi } _ 0}{h}r]|_{\bar{r}+},\end{gathered}\ ] ] and the traction - free boundary condition at @xmath345 @xmath346|_{y=1}=0.\end{aligned}\ ] ] for the part of the plate without piezoceramic patches ( @xmath347 ) the solution is given by @xmath348 while for the part covered by the piezoceramic patches ( @xmath349 ) we have @xmath350 here @xmath351 and @xmath352 are bessel function of the first and second kind , respectively .
the constants @xmath353 can be determined from the jump conditions at @xmath343 and the boundary condition at @xmath345 . substituting the formulas for @xmath354 into them , we get the system of linear equations @xmath355 where @xmath356 and @xmath357 are given by @xmath358 after finding @xmath359 we can determine the amplitude of @xmath360 by the formula @xmath361,\end{aligned}\ ] ] where @xmath362
. then the amplitude of the total charge on one of the electrodes is equal to @xmath363\ } , \end{aligned}\ ] ] where @xmath364 is the solution of the system @xmath365 with @xmath366 according to the solution of this problem the resonant frequencies are the roots of the determinantal equation @xmath367 the antiresonant frequencies should be found from the condition that the total charge vanishes giving @xmath368 + \bar{a}_3[y_1(\vartheta ) -\bar{r}y_1(\vartheta \bar{r } ) ] .
\label{7.7}\ ] ] .material constants of pzt-5a [ cols="<,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] for the numerical simulations we use as an example aluminum oxide ( al@xmath369o@xmath370 ) as the isotropic elastic ( dielectric and non - conducting ) material and pzt-5a polarized in the third direction as the piezoceramic material .
the material constants for pzt-5a , taken from @xcite , are presented in table .
[ table1 ] , with @xmath371f / m being the dielectric constant of vacuum . in this table
the dimension unit for @xmath372 is @xmath373m@xmath374/n , for @xmath375 is @xmath373c / n , and for @xmath376 is @xmath377kg / m@xmath378 . besides , for piezoceramic material we have @xcite @xmath379 where @xmath380 is poisson s ratio and @xmath381 the planar coupling factor . from table .
[ table1 ] we find that @xmath382 and @xmath383 . of the uncovered area at fixed @xmath384.,height=264
] the material constants of aluminum oxide , taken from @xcite , are @xmath385 furthermore , for an elastic ( dielectric ) isotropic material @xmath386 at fixed @xmath387.,height=264 ] with these material data we can compute the numerical values of constants used in this 2-d problem .
the results of numerical simulations are shown in figures 2 , 3 , and 4 .
figure 2 plots the first three roots of equations as functions of the parameter @xmath342 for @xmath384 representing the resonant frequencies of the smart sandwich plate versus @xmath388 .
figure 3 shows the first three roots of equations as functions of the parameter @xmath330 changing between 0.1 and 0.5 at fixed @xmath387 representing the resonant frequencies versus @xmath330 .
it is interesting to note that , in the limit @xmath389 and @xmath390 , the resonant frequencies tend to those of the piezoceramic homogeneous plate covered by the electrodes , while in the limit @xmath391 and @xmath392 they tend to those of the purely elastic plate . at fixed @xmath393.,height=264 ] concerning the anti - resonant frequencies we observe that they lye slightly above the resonant ones only if @xmath330 is close to zero as show in fig . 4 for @xmath393 , where the resonant frequencies correspond to the bold lines , while the anti - resonant frequencies to the dashed line .
this is close to the behavior of the purely piezoceramic plate @xcite .
when @xmath330 is close to 1/2 which means that the piezoceramic patches are thin compared to the thickness of the elastic layer , the anti - resonant frequencies are much higher than the resonant ones .
thus , the sandwich plate with thin piezoceramic patches does not have electrically anti - resonant vibrations near the resonant frequencies .
it is shown in this paper that the rigorous first order approximate 2-d theory of thin smart sandwich shells can be derived from the exact 3-d piezoelectricity theory by the variational - asymptotic method .
the strains and electric field of the smart sandwich shell turn out to be discontinuous through the thickness and differ essentially from those of the homogeneous piezoelectric shells . the error estimation for the constructed 2-d theory is established that enables one to apply this theory to the problem of vibration control of thin elastic shells with bonded piezoelectric patches .
note that the elastic layer of smart structures used in engineering praxis ( like epoxy / glass material ) may have more complicated anisotropy property for which the theory developed here is not applicable .
our study in such cases is still going on and the results will be reported elsewhere .
cheng , z.q . ,
lim , c.w . ,
kitipornchai , s. ( 2000 ) . three - dimensional asymptotic approach to inhomogeneous and laminated piezoelectric plates . international journal of solids and structures 37(23 ) , 3153 - 3175 .
he , x.q . ,
liew , k.m .
, ng , t.y . , sivashanker , s. ( 2002 ) . a fem model for the active control of curved fgm shells using piezoelectric sensor / actuator layers .
international journal for numerical methods in engineering 54(6 ) , 853 - 870 .
lee , c.k .
theory of laminated piezoelectric plates for the design of distributed sensors / actuators .
part i : governing equations and reciprocal relationships .
the journal of the acoustical society of america 87(3 ) , 1144 - 1158 .
touloukian , y.s .
thermophysical properties of high temperature solid materials .
oxides and their solutions and mixtures .
thermophysical and electronic properties information analysis center lafayette in .
tzou , h. s. , gadre , m. ( 1989 ) . theoretical analysis of a multi - layered thin shell coupled with piezoelectric shell actuators for distributed vibration controls .
journal of sound and vibration 132(3 ) , 433 - 450 .
wein , f. , kaltenbacher , m. , bnsch , e. , leugering , g. , schury , f. ( 2009 ) .
topology optimization of a piezoelectric - mechanical actuator with single - and multiple - frequency excitation .
international journal of applied electromagnetics and mechanics 30(3 , 4 ) , 201 - 221 . | an asymptotically exact two - dimensional theory of elastic - piezoceramic sandwich shells is derived by the variational - asymptotic method .
the error estimation of the constructed theory is given in the energetic norm .
as an application , analytical solution to the problem of forced vibration of a circular elastic plate partially covered by two piezoceramic patches with thickness polarization excited by a harmonic voltage is found . elastic , piezoelectric , sandwich , shell , variational - asymptotic method . |
the production of heavy flavours ( charm and beauty ) in high energy @xmath0 collisions at provides new opportunities to study the dynamics of perturbative qcd and to extract information on the proton and photon structure .
the heavy - flavour cross section at is dominated by photoproduction events where the electron is scattered by a small angle , producing photons of almost zero virtuality .
the dynamics of charm and beauty photoproduction can be probed at in a wide kinematical region , the available centre - of - mass energy being more than one order of magnitude higher than in fixed - target experiments .
resolved - photon interactions , where the photon fluctuates into parton constituents , which undergo hard scattering with the partons from the proton , contribute significantly at energies and allow one to study the photon structure . because of the high cross sections in photoproduction , some sort of tagging must be done to allow the data to be recorded to tape in the experiment .
a charm meson may be tagged and a jet finder may also be used , with a certain number of high transverse energy jets demanded . in the first case information on the charm meson is available . in the second case the observables of the meson , the jet and the correlation between the two may be examined .
this sheer wealth of variables and currently available statistics is what makes photoproduction most interesting .
all the previous studies from normal photoproduction may be repeated with a charm tag , but thought should be given to original analyses . as a complement , heavy - flavour production at can be studied in deep inelastic scattering ( dis ) .
the finding of a scattered positron and the distribution of these events in bins of the parton momentum fraction @xmath1 and the photon virtuality @xmath2 is well documented in the literature . with the additional requirement of a tagged charm meson ( typically @xmath3 ) a way is found to measure
the charm tagged content of the proton structure function @xmath4 ( @xmath5 ) .
since @xmath5 is evolved from the gluon content , it provides a measurement of the gluon density in the proton .
large statistics in the normal @xmath4 measurement and the added measurement of @xmath5 constrain the theory greatly .
inclusive @xmath4 is predicted to contain between 10% and 25% charm and it must therefore be properly taken into account in global analyses of structure function data .
differential cross sections for @xmath3 production in dis have also become available and allow a more detailed study of the charm production mechanisms than is possible with the inclusive @xmath5 measurement alone .
the production of heavy - quark bound states has been the subject of intense study during the past few years .
exciting phenomenological developments followed from the application of non - relativistic qcd ( nrqcd ) , an effective theory that disentangles physics on the scale of the heavy - quark mass , relevant to the production of a heavy - quark pair , from physics on the scale of the bound state s binding energy , relevant to the formation of the quarkonium .
the nrqcd approach implies that so - called colour - octet processes , in which the heavy - quark antiquark pair is produced at short distances in a colour - octet state and subsequently evolves non - perturbatively into a physical quarkonium , must contribute to the cross section .
although gluon fragmentation into colour - octet quark pairs appears as the most plausible explanation of the large direct @xmath6 production cross section observed at the tevatron , nrqcd factorization is still not definitely established on a quantitative level .
the analyses of inclusive charmonium production at offer unique possibilities to test general features of the quarkonium production mechanisms and to establish the phenomenological significance of colour - octet processes . in the following
we will discuss various aspects of charm and beauty production at , organized into the three above - mentioned areas : photoproduction , deep inelastic scattering and quarkonium production .
additional information may be found in .
the photon structure is being probed in new kinematic regimes at hera .
tagging of charmed mesons or semileptonic decays in photoproduction events is providing a wealth of information on charm in the photon and the processes involved in its production .
thanks to the overall higher cross section for photoproduction events , there exists a large statistics sample from the 1996 and 1997 data sets alone ( 37 pb@xmath7 ) . since photoproduction of charm and beauty is much less affected by non - perturbative and higher - order effects than hadroproduction cross sections , the measurements can provide important tests of the heavy - flavour production dynamics . the photon
proton cross section may be written as a sum of a direct and a resolved - photon contribution : @xmath8 where @xmath9 and @xmath10 denote the density functions for parton @xmath11 and @xmath12 in the proton and in the photon , respectively .
the renormalization scale is labelled @xmath13 , and @xmath2 is the hard scale of the interaction .
the factorization scales @xmath14 and @xmath15 are related to the subtraction of divergences associated with the collinear emission of partons from the incoming parton in the proton and in the incoming photon , respectively .
the factorization of collinear initial - state singularities has to be performed at nlo where feynman diagrams like figure [ factor ] contribute .
although this diagram forms part of the @xmath16 cross section , if the virtuality @xmath17 of the propagator labelled in the figure is less than @xmath18 , then its contribution is already included in the lo - resolved process .
hence the separation of direct and resolved reactions is ambiguous beyond lo and only the sum is meaningful .
this is explicitly shown in ( [ dir - res ] ) , where the factorization scale @xmath15 appears in both the direct and the resolved contributions .
total cross sections , single - inclusive distributions as well as heavy - quark correlations have been calculated to next - to - leading order accuracy @xcite . in higher orders , potentially large terms @xmath19 arise from the collinear branching of gluons or photons into heavy - quark pairs or from the collinear emission of gluons by a heavy quark at large transverse momentum .
for @xmath20 these terms may spoil the convergence of the perturbation series and should be resummed by absorbing the collinear @xmath21-singularities into heavy - quark parton densities and fragmentation functions .
applications of this approach to charm and @xmath3 photoproduction at @xmath22 can be found in @xcite . for a comparison of the different theoretical schemes ,
see @xcite .
experimentally , the first topic to be studied , in general , is the production of charmed mesons , which may be identified by their decay products .
both h1 and zeus have shown results on the @xmath3 meson in photoproduction @xcite .
this channel is favoured because of its clear signal .
a comparison of data and different nlo theory predictions for the @xmath3 pseudorapidity distribution is shown in figure [ php_eta ] .
the data lie above the nlo theory when the @xmath3 meson travels in the forward direction , as is also seen in the dis case .
although the discrepancy might not be considered very significant at present , some interesting suggestions were made for its possible causes .
they are listed here and discussed below : 1 . `
string effects ' between the proton remnant and the @xmath23 quark @xcite .
low-@xmath24 non - perturbative gluon in the photon @xcite
. this would give rise to more charm and is currently poorly constrained by lep data .
effects attributed to the peterson et al . fragmentation function used .
4 . more beauty than predicted by the theory . item ( i ) can be examined by comparing the available lo monte carlo programs to the data . in the fragmentation process a string would be linked between the charm quark and the forward travelling proton remnant .
the tension in this string could be responsible for pulling the quark ( and hence the meson ) forward .
this effect is not implemented in the nlo calculations , so lo monte carlo would be more forward than nlo calculations .
this comparison was carried out at the recent hera monte carlo workshop @xcite and a better agreement is seen between the shape of the lo curves and the data ( see figure [ eta_xs ] ) . however , there are many other effects implemented in the monte carlo ( including item ( iii ) ) .
the process of separating them is still ongoing .
previously zeus demonstrated @xcite its sensitivity to the low-@xmath24 regions , by the use of the observable @xmath25 where the sum is over the two jets with highest transverse energy . in lo
, @xmath26 can be interpreted as an observable related to the fraction of the photon s momentum participating in the hard - scale interaction . at higher
@xmath26 ( the usual cut is 0.75 ) the direct processes dominate , and at lower @xmath27 the resolved processes dominate .
if a charm meson or semileptonic decay is then tagged in these events , statements can be made about the charm component of the photon .
this is of interest because charm in the photon may come from the low-@xmath24 non - perturbative gluon , a parton density of the photon still badly constrained by lep @xcite .
demanding a jet also provides another , even harder , scale to the calculation .
typical experimental jet transverse energies are of the order of 6 or 7 gev , well above the charm mass .
when @xmath28 , charm should behave as a light quark . at the scales probed with present data it is unclear to what extent this is true , and the problem arises of how to treat the intermediate @xmath2 region .
similar theoretical issues arise for charm production in dis and will be discussed in the next section .
zeus have published a measurement of @xmath26 @xcite with a tagged @xmath3 meson , shown in figure [ xgamma_dstar ] .
this shows a clear tail to low @xmath26 , which in terms of lo monte carlo programs can only be explained by a large ( 40% ) component of lo - resolved processes .
this is heavily dependent on the amount of charm in the parton density function used .
a nlo calculation , performed in a factorization scheme where heavy quarks are exclusively generated at short distances from light - parton scattering , can not describe this long tail and falls short of both the data and the lo monte carlo .
this could be due to a large non - perturbative part of the gluon in the photon , or could be attributable to effects from the fragmentation function used .
heavy - hadron fragmentation functions were discussed as an interesting area of study and not just as a possible explanation as item ( iii ) on the above list .
the extraction of fragmentation functions from experimental data and the value of the corresponding non - perturbative parameters are closely related to the description of the perturbative part of the production cross section .
different theoretical schemes for calculating heavy - quark production require different heavy - hadron fragmentation functions .
( see @xcite for recent nlo extractions of @xmath3 fragmentation functions from @xmath29 data . )
suggestions have been made for defining and directly measuring fragmentation functions at hera @xcite .
however , the invariant mass of the subprocess can not be determined directly and has to be reconstructed from observables in the final state ( e.g. jet variables ) .
it therefore makes sense to define the fragmentation parameter in terms of the jets .
one example would be : @xmath30 where @xmath31 is the four momentum of the proton and @xmath32 and @xmath33 are those of the meson and the jet , respectively . a modified fragmentation function
may then be defined as @xmath34 only a convolution of the @xmath35 and @xmath36 fragmentation functions may be measured , so direct comparison with the lep data is impossible . with suitably large statistics ,
it may be possible to do a search for scaling violations with hera data alone .
work is now proceeding on the measurement of charm fragmentation functions at hera .
finally , the subject of open beauty production was brought up .
h1 already had released their preliminary measurement and comparison with the lo theory @xcite .
h1 quote , for the kinematic range @xmath37 , @xmath38 , @xmath39 gev and @xmath40 , the preliminary visible cross section of @xmath41 nb and also quote a predicted cross section from aroma 2.2 @xcite of @xmath42 nb .
this states that the preliminary measurement of the cross section of open beauty was up to a factor of 5 higher than the lo theory as predicted by the monte carlo aroma .
it should be noted that aroma is designed to produce only a direct component .
the magnitude of the resolved component is not taken into account in the comparison . at the time of the workshop ,
no result from zeus was forthcoming .
however , before going to press , a preliminary result was released of a beauty measurement by the study of semileptonic decays to electrons @xcite .
the measurement is an inclusive electron measurement @xmath43 in the kinematic region : * @xmath37 , @xmath44 . * two jets with @xmath45 gev and @xmath46 . * an electron in the final state with @xmath47 gev and @xmath48 .
this measurement is inclusive of all beauty , charm and light - quark decays , but with all detector effects removed ( e.g. conversion electrons ) .
the @xmath26 distribution ( figure [ xg - sl ] ) shows , as with the @xmath3 study , a clear tail to low @xmath27 , which needs a significant fraction of resolved lo monte carlo to describe it .
the @xmath26 distribution also peaks at high values , which is consistent with the observation of direct processes .
the preliminary fit on the fraction of the lo resolved contribution yields @xmath49% and agrees well with the herwig 5.9 @xcite prediction of 40% .
the agreement in shape between the data and the lo monte carlo is good .
next the @xmath50 distribution was examined . in this study
@xmath50 is the momentum of the electron transverse to the jet axis of the closest jet ( the jets were reconstructed using a clustering algorithm , ktclus @xcite ) .
the electrons from beauty decays are expected to dominate at high @xmath50 because of the larger beauty mass .
the obtained distribution is shown in figure [ pt - sl ] .
a fit was done allowing the fraction of beauty to vary with respect to the charm and light quarks contribution ( added in the ratios given by herwig ) . a preliminary beauty fraction of @xmath51% is needed to fit the data ( using the 40% resolved component ) , in agreement with the herwig prediction of @xmath52% .
zeus quotes a preliminary visible cross section for beauty production of @xmath53 in the kinematic region described above .
finally , a preliminary overall normalization factor of 3.7 is necessary to fit the monte carlo to the data .
this normalization factor is for the complete sample and not just for the beauty component .
it should be noted that these results depend on the parton density functions used in the analysis ( for the photon grv - lo @xcite and for the proton cteq4 @xcite ) and also on the charm and beauty masses used ( @xmath54 gev and @xmath55 gev ) .
it should also be noted that for normal dijet photoproduction a factor of 1.8 is generally needed to normalize the monte carlo to the data . in the @xmath3 with dijets study , a factor of 2.6 was necessary .
there is clear consistency between herwig 5.9 and the zeus data in the percentages and @xmath50-distribution shapes .
however , they disagree with the large normalization factor . taking all the caveats in the previous paragraph into account and the different measured kinematic regimes , it is hard to make a statement about the agreement between the experiments at present .
a comparison of the visible cross section with a nlo calculation as well as alternative experimental techniques using microvertex detection should allow a clarification of the issue in the future .
the production of charm quarks in has become an important theoretical and phenomenological issue . the charm contribution to the total structure function
@xmath4 at small @xmath1 , at hera , is sizeable , up to @xmath56 ; through the charm contribution to the scaling violations the treatment of charm also has a significant impact on the interpretation of the nmc data . besides being an interesting theoretical problem due to the presence of two hard scales , @xmath57 and @xmath58 ,
a proper description of charm - quark production in dis is thus required for a global analysis of structure function data and a precise extraction of the parton densities in the proton . at scales
@xmath59 , charm production in is calculated in the so - called fixed flavour number scheme ( ffns ) from hard processes initiated by light quarks ( @xmath60 ) and gluons , where all effects of the charm quark are contained in the perturbative coefficient functions .
the ffns incorporates the correct threshold behaviour , but for large scales , @xmath61 , the coefficient functions in the ffns at higher orders in @xmath62 contain potentially large logarithms @xmath63 , which may need to be resummed .
such a resummation can be achieved by including the heavy quark as an active parton in the proton .
the simplest approach incorporating this idea is the so - called zero mass variable flavour number scheme ( zm - vfns ) , where heavy quarks are treated as infinitely massive below some scale @xmath64 and massless above this threshold .
this scheme has been used in global fits for many years , but it has an error of @xmath65 and is not suited for quantitative analyses unless @xmath66 .
considerable effort has been made to devise a scheme for heavy - flavour production that interpolates between the ffns close to threshold and the zm - vfns at large @xmath58 . here , we will focus on two such schemes , the aivaziz collins olness tung ( acot ) @xcite and the thorne
roberts ( tr ) @xcite scheme , which have been used in recent global analyses of parton distributions .
these generalized vfnss include the heavy quark as an active parton flavour and involve matching between the ffns with three active flavours and a four - flavour scheme with non - zero heavy - quark mass .
they employ the fact that the mass singularities associated with the heavy - quark mass can be resummed into the parton distributions without taking the limit @xmath67 in the short - distance coefficient functions , as done in the zm - vfns . to all orders ( and neglecting intrinsic heavy - quark contributions @xcite ) the ffns , the acot scheme , and the tr scheme are identical , but
the way of ordering the perturbative expansion is not unique and the results differ at finite order in perturbation theory .
a nlo comparison of the predictions for @xmath68 within the ffns @xcite , the zm - vfns and the tr scheme is shown in figure [ dis - schemes ] . while the zm - vfns does not provide an adequate theoretical description at low values of @xmath58 , the difference between the ffns and the tr - scheme is moderate , typically @xmath69 .
a similar behaviour is observed for the acot - vfns .
experimentally , zeus have released their @xmath5 measurement using their full 1996 and 1997 data samples @xcite .
h1 have released their measurement using only the 1994 to 1996 data sample @xcite .
figure [ f2cexp ] shows the @xmath2 and @xmath1 dependence of @xmath68 compared to the nlo ffns calculation @xcite .
the measurements are in good agreement with the theoretical prediction . given the experimental error and the theoretical uncertainty due to the variation of the charm - quark mass , as shown in figure [ f2cexp ] , present data on @xmath68 can not discriminate between the ffns and the acot and tr schemes .
global structure function fits , however , seem to favour the vfnss over the ffns @xcite .
the resummation of @xmath63 terms at higher orders in the ffns may not be necessary for the present analysis of neutral - current structure functions , but it is required for a stable theoretical prediction of the charged - current structure function @xmath70 @xcite . as analysed in @xcite , the higher - order corrections to the lowest - order flavour excitation mechanism @xmath71 are large and dominated by @xmath72-boson
gluon fusion @xmath73 at @xmath74 .
the extraction of the strange quark density from the charged - current structure functions at small @xmath1 thus requires a good knowledge of the gluon density .
another source of potentially large higher - order corrections to charm production in dis comes from the emission of soft gluons in the threshold region .
this effect has been studied within the ffns for charm production through photon
gluon fusion in @xcite .
the fact that heavy - quark production at hera energies may be sensitive to threshold effects is due to the large gluon density at small @xmath1 , which enhances the contribution of soft - gluon emission near the elastic limit .
an all - order resummation of threshold logarithms has been performed in @xcite at next - to - leading logarithmic accuracy ; an approximate nnlo result has been derived for @xmath68 and the single - particle inclusive distribution @xmath75 .
figure [ dis - nnlo ] shows the effect of the approximate nnlo corrections for the inclusive structure function .
plotted are the @xmath76-factors @xmath77 ( solid line ) and @xmath78 ( dashed line ) .
it can be concluded that the soft - gluon effects are well under control for @xmath79 , where the size of the nnlo corrections , evaluated at the central scale @xmath80 , is negligible .
it has also been shown in @xcite that the inclusion of nnlo terms stabilizes the theoretical prediction for @xmath68 by reducing the scale dependence of the nlo result .
charm production in dis has not only been studied for inclusive structure functions but differential cross sections for @xmath3 production are available as well @xcite .
differential distributions allow a study of the underlying production mechanism more detailed than what is possible with the total cross section alone .
a comparison of recent data for differential @xmath3 cross sections in dis with the nlo ffns calculation @xcite shows good overall agreement , with a possible slight excess of data in the forward region for large @xmath81 and in the central @xmath82 distribution ( see figure [ dstar_diff ] ) ( @xmath82 is the fractional momentum of the @xmath3 in the @xmath83 frame ) .
we have verified that the excess is not driven by the small-@xmath58 region of the data , so a possible resolved photon contribution can be ruled out .
hadronization effects may well be responsible for the deviation from the parton - level calculation , as discussed for @xmath3 photoproduction in the previous section .
recently , new theoretical results for differential distributions in the acot scheme have become available @xcite .
an event generator monte carlo program has been constructed , which allows to produce arbitrary differential distributions and to impose experimental cuts .
figure [ acot - pt ] shows a comparison between the acot prediction for the @xmath84 distribution in @xmath3 dis and recent data @xcite .
given the uncertainty in the choice of the non - perturbative @xmath3 fragmentation function , which affects the @xmath84 slope of the theory , the agreement is good . in figure [ acot - pt ]
the peterson et al .
form @xcite has been adopted , with @xmath85 .
a smaller value for @xmath86 , as suggested by recent analyses of @xmath29 data @xcite , would increase the cross section at large @xmath84 and thereby improve the theoretical description of the data further .
a more detailed comparison between data and vfns predictions will be carried out in the future .
experimentally much effort is now being put into reducing the systematic and statistical errors , to gain as much as possible from the current data samples .
first , detailed systematic studies are in progress for standard @xmath4 , and these will eventually be applied to the measurement of @xmath5 .
some of the systematic error on @xmath5 is because of the extrapolation , due to the forward excess seen by both collaborations @xcite .
work is under way to implement ` string effects ' described in the previous section into nlo calculations .
these appear to correct the calculation so as to agree with the data in the forward direction .
the statistical errors may not be much improved by 1998 and 1999 data taking , so work is under way to examine different channels . already zeus have shown a tagged semileptonic decay channel @xcite , and other channels are under investigation ( e.g. @xmath87 ) .
since these are statistically independent samples , it may be possible to combine them to reduce the statistical errors .
finally it is hoped that the new hera luminosity upgrade planned for next year and the inclusion of results from microvertex detectors into the analyses will provide vastly increased statistics as well .
the production of quarkonium at high - energy colliders has been the subject of considerable interest during the past few years .
new experimental results from @xmath88 , @xmath0 and @xmath29 colliders have become available , some of which revealed dramatic shortcomings of earlier quarkonium models ( see @xcite for recent reviews ) . in theory ,
progress on factorization between perturbative and the quarkonium bound state dynamics has been made .
the ` colour - singlet model ' ( csm ) has been superseded by a consistent and rigorous approach , based on non - relativistic qcd ( nrqcd ) @xcite , an effective field theory that includes the so - called colour - octet mechanisms .
on the other hand , the ` colour evaporation ' model ( cem ) @xcite of the early days of quarkonium physics has been revived @xcite . despite these developments ,
the range of applicability of these approaches to the practical case of charmonium is still subject to debate , as is the quantitative verification of factorization .
the problematic aspect is that , because the charmonium mass is still not very large with respect to the qcd scale , non - factorizable corrections may not be suppressed enough , if the quarkonium is not part of an isolated jet , and the expansions in nrqcd may not converge very well . in this
situation cross checks between various processes , and predictions of observables such as quarkonium polarization and differential cross sections , are crucial in order to assess the importance of different quarkonium production mechanisms , as well as the limitations of a particular theoretical approach . according to the nrqcd formalism , inclusive quarkonium production can be viewed as a two - step process , where the production of a heavy - quark pair @xmath89 $ ] in a certain angular momentum and colour state @xmath90 is described by a perturbatively calculable short - distance cross section @xmath91 and the subsequent quarkonium formation is parametrized by a non - perturbative matrix element @xmath92 .
the magnitude of the non - perturbative transition probabilities is determined by the intrinsic velocity @xmath93 of the bound state and the production cross section can be expressed as a double expansion in @xmath62 and @xmath93 . in the cem and
the related ` soft colour interaction ' ( sci ) model @xcite , the cross section for a specific charmonium state is given as a fraction of the inclusive @xmath94 production cross section integrated up to the open charm threshold .
the production fraction is assumed to be universal in the cem , while in the sci model it depends somewhat on the partonic state and the possible string configurations . in both models
the colour and spin quantum numbers of the intermediate @xmath95 state are irrelevant , and gluon emission during hadronization is assumed to be unsuppressed .
the analysis of charmonium production in high energy @xmath0 collisions at hera appears to be a powerful tool to test the nrqcd approach and to discriminate nrqcd from the cem or sci model .
the production of charmonium at is dominated by photon
gluon fusion , where the photon interacts directly with a gluon from the proton .
resolved processes are expected to contribute significantly to the lower end - point of the @xmath96 energy spectrum and might be probed at for the first time in the near future .
higher - twist phenomena such as diffractive charmonium production @xcite are not included in the leading - twist calculations of nrqcd and the cem : it can be eliminated by either a cut in the @xmath96 transverse momentum @xmath97@xmath98 gev , or by a cut in the @xmath96 energy variable @xmath99 ( in the proton rest frame , @xmath100 is the ratio of the @xmath96 to @xmath101 energy , @xmath102 ) . within nrqcd and at leading order in the intrinsic velocity @xmath93 of the bound state ,
the inelastic photoproduction of @xmath96 comes from an intermediate @xmath95 pair in a colour - singlet @xmath103 state and coincides with the colour - singlet model result .
( the notation for the angular momentum configuration is @xmath104 with @xmath105 , @xmath106 and @xmath107 denoting spin , orbital and total angular momentum , respectively . )
cross sections @xcite , polar @xcite , and polar and azimuthal @xcite decay angular distributions have been calculated for the direct - photon contribution .
the angular integrated cross section is known to next - to - leading order in @xmath62 @xcite .
relativistic corrections to the colour - singlet channel at @xmath108 were calculated in @xcite and were found to be small in the inelastic region @xmath109 , @xmath110 gev .
correction in the large-@xmath100 region even after a cut in the transverse momentum , @xmath111 gev , has been applied . ]
the colour - singlet contribution , including next - to - leading corrections in @xmath62 , is known to reproduce the current @xmath96 photoproduction data @xcite adequately .
but there is still a considerable amount of uncertainty in the normalization of the theoretical prediction , which arises from the value of the charm quark mass and the wave function at the origin , as well as the choice of parton distribution functions and renormalization / factorization scale . at order @xmath112@xmath113 relative to the colour - singlet contribution
, the @xmath96 can also be produced through the intermediate colour - octet configurations @xmath103 , @xmath114 and @xmath115 . in the inelastic region , they have been considered in @xcite for the direct photon contribution and in @xcite for resolved photons , in which case the photon participates in the hard scattering through its parton content .
colour - octet contributions to the total photoproduction cross section ( integrated over all @xmath100 and @xmath84 ) are known to next - to - leading order @xcite .
the polarization of inelastically produced @xmath96 due to these additional production mechanisms has been calculated recently @xcite .
the colour - octet production channels are kinematically different from the colour - singlet one , because the @xmath114 and @xmath115 configurations can be produced through @xmath116-channel exchange of a gluon already at lowest order in @xmath62 .
( for the @xmath103 octet this is true for the resolved process . )
this leads to a significantly enhanced amplitude , in particular in the large-@xmath100 region .
the colour - octet contributions to @xmath96 photoproduction are indeed strongly peaked at large @xmath100 @xcite .
such a shape is not supported by the data , which at first sight could lead to a rather stringent constraint on the octet matrix elements @xmath117\rangle$ ] , @xmath118 , and to an inconsistency with the values obtained for these matrix elements from other processes . however , it is premature to interpret the discrepancy between the theoretical predictions for colour - octet contributions and the data at large @xmath100 as a failure of the nrqcd approach itself , as long as not all theoretical uncertainties that reside in the calculation of the short - distance cross sections have been assessed carefully .
potentially large qcd effects include higher - twist contributions @xcite , higher - order corrections and intrinsic transverse momentum of the partons in the proton @xcite .
moreover , the peaked shape of the @xmath100-distribution is derived neglecting the energy transfer in the non - perturbative transition @xmath119\to j/\psi+x$ ] . in reality
the peak may be considerably smeared @xcite as a consequence of resumming kinematically enhanced higher - order corrections in @xmath120 and no constraint or inconsistency can be derived from the end - point behaviour of the @xmath100-distribution at present . as a consequence ,
the role of octet contributions to the direct process remains unclear .
the resolved photon contribution , on the other hand , could be entirely colour - octet - dominated @xcite .
the @xmath100-distribution should then begin to rise again at small @xmath100 , if the colour - octet matrix elements are as large as suggested by nrqcd velocity scaling rules @xcite and fits to hadroproduction data .
the cem and the sci model also predict a strong colour - octet enhancement at small and large values of @xmath100 , again due to @xmath116-channel gluon - exchange diagrams @xcite . as in the case of the nrqcd analysis
, the impact of qcd effects ( higher - twist contributions , higher - order corrections and intrinsic transverse momentum ) has to be analysed before a final conclusion can be drawn .
since the qualitative behaviour of the @xmath96 energy distribution is similar in all existing approaches that include colour - octet contributions , it is not well suited to discriminate between nrqcd , cem and sci .
the most powerful way to test nrqcd against the cem / sci is the analysis of charmonium polarization .
the polarization analysis in nrqcd @xcite is based on the symmetries of the nrqcd lagrangian , of which spin and rotational symmetry are crucial , and the charmonium polarization can be calculated from the spin orientation of the intermediate @xmath95 state .
in contrast to the nrqcd approach , the cem / sci predict charmonium to be produced always unpolarized .
the models assume unsuppressed gluon emission from the @xmath95 pair during hadronization , which randomizes spin and colour .
this assumption is wrong in the heavy - quark limit where spin symmetry is at work and soft gluon emission does not flip the heavy - quark spin . nonetheless , since the charm - quark mass is not very large with respect to the qcd scale , the applicability of heavy - quark spin symmetry to charmonium physics has to be tested by confronting the nrqcd predictions to experimental data .
two instructive examples for @xmath96 photoproduction at are shown in figure [ qq - pol ] @xcite , namely the polar - angle parameter @xmath121 and the azimuthal - angle parameter @xmath122 in the leptonic @xmath96 decay , as a function of the energy fraction @xmath100 and the transverse momentum @xmath84 , respectively .
the polarization signatures predicted from colour - octet processes within nrqcd are distinctive and can also be used to extract information on the relative weight of colour - singlet and colour - octet contributions .
polarization signatures have also been studied for @xmath123 production in @xmath124 collisions at @xcite .
nrqcd predicts the @xmath123 mesons to be produced transversely polarized , again in contrast to the cem / sci , which imply unpolarized quarkonium .
another powerful way of discriminating nrqcd against the cem is the measurement of @xmath125 photoproduction . in nrqcd
the @xmath126 ratio is process - dependent and strongly suppressed in photoproduction with respect to hadroproduction . on the other hand
, the assumption of a single universal long - distance factor in the cem implies a universal @xmath126 ratio , which is not , however , supported by the comparison of charmonium production in fixed - target hadron collisions and early photoproduction experiments
. a search for @xmath125 production at , resulting in a cross section measurement or upper cross section limit would be crucial to settle this issue .
recently , first experimental results on inclusive @xmath96 electroproduction have been presented @xcite and compared to a leading - order nrqcd calculation including colour - singlet as well as colour - octet contributions @xcite . at large photon virtualities ,
the analysis of @xmath96 leptoproduction is under better theoretical control than photoproduction , since higher - order corrections and higher - twist contributions should become less important as @xmath2 increases .
also , at sufficiently large @xmath2 , diffractive contributions are expected to be suppressed .
the experimental data for differential cross sections in the kinematic region @xmath127
gev@xmath128 are shown in figure [ qq - elec ] , together with the nrqcd predictions .
the colour - singlet channel underestimates the data by a factor @xmath129@xmath130 , but there still is a considerable amount of uncertainty in the normalization from the value of the charm - quark mass , the wave - function at the origin , higher - order corrections as well as the choice of parton distribution functions and renormalization / factorization scale . the normalization of the theoretical prediction is significantly improved when adding colour - octet contributions ; the differential distributions , however , seem to favour the shape predicted by the colour - singlet cross section . more theoretical work and more experimental information extending to larger @xmath2 values
are needed before firm conclusions can be drawn .
various other observables and final states can be accessed with higher - statistics data , like photoproduction of @xmath131 and @xmath123 particles , associated @xmath132 production @xcite as well as fragmentation contributions at large @xmath84 @xcite .
the future analyses at will offer unique possibilities to test the nrqcd approach and to study the mechanism of quarkonium production in general .
we wish to thank the organizers of the 3rd uk phenomenology workshop on hera physics for creating a stimulating and pleasant atmosphere and the heavy - flavour plenary speakers jo cole , giovanni ridolfi and robert thorne for their input and enthusiasm .
we would also like to thank matthew wing and mark sutton , without whom certain results would not have made these proceedings .
we are grateful to jim amundson , matteo cacciari , rohini godbole , susanne mohrdieck , fred olness , dick roberts , dave soper and wu - ki tung for valuable discussions and for providing us with results for presentation in this summary .
finally , many thanks go to all the participants of the workshop who have contributed with presentations and discussions .
999 p. nason , s. dawson , r.k .
ellis , nucl .
* b303 * ( 1988 ) 607 ; g. altarelli , m. diemoz , g. martinelli , p. nason , nucl .
* b308 * ( 1988 ) 724 ; w. beenakker , h. kuijf , w. van neerven , j. smith , phys .
rev . * d40 * ( 1989 ) 54 ; r.k .
ellis , p. nason , nucl .
* b312 * ( 1989 ) 551 ; m.l .
mangano , p. nason , g. ridolfi , nucl .
* b373 * ( 1992 ) 295 ; j. smith , w.l .
van neerven , nucl .
phys . * b374 * ( 1992 ) 36 ; s. frixione , m.l . mangano , p. nason , g. ridolfi , nucl .
phys . * b412 * ( 1994 ) 225 and cern - th/97 - 16 , february 1997 , to be published in _ heavy flavours ii _ , ed . by a.j .
buras , m. lindner , world scientific ( hep - ph/9702287 ) .
l. lnnblad , on behalf of the lund group , in discussions , and : e. norrbin , t. sjstrand , phys . lett .
* b442 * ( 1998 ) 407 ; e. norrbin , lu tp 98 - 24 , december 1998 , to appear in the proceedings of the workshop on heavy quarks at fixed target , ed . by h.w.k .
cheung , j.n . butler .
j. amundson , f. olness , c. schmidt , w.k .
tung , x. wang , fermilab - conf-98 - 153-t , july 1998 , to be published in the proceedings of 6th international workshop on deep inelastic scattering and qcd ( dis 98 ) , brussels , belgium , 48 april 1998 ; j. amundson , c. schmidt , w.k .
tung , x. wang , work in progress .
r. gavai et al . , int .
j. mod .
* a10 * ( 1995 ) 3043 ; j.f .
amundson , o.j.p .
boli , e.m . gregores , f. halzen , phys . lett .
* b372 * ( 1996 ) 127 ; phys .
lett . * b390 * ( 1997 ) 323 ; g.a .
schuler , r. vogt , phys .
* b387 * ( 1996 ) 181 ; o.j.p .
eboli , e.m .
gregores , f. halzen , madph-98 - 1045 ( hep - ph/9802421 ) ; a. edin , g. ingelman , j. rathsman , phys .
* d56 * ( 1997 ) 7317 . | we review the theoretical and experimental status of heavy - flavour production at hera .
the results presented include some outstanding issues in charm and beauty photoproduction , charm production in dis and quarkonium production .
cern - th/99 - 30 + ucl / hep 99 - 03 + january 1999 |
the evolution of galaxies and of the intergalactic / intracluster medium ( igm / icm ) are closely connected , via the continuous interchange between the ism and the igm / icm .
the heating and metal enrichment of the igm / icm , and the enrichment of galaxies ism with pristine gas via infall from the igm are manifestations of this close connection .
evidence for gas outflows from galaxies has been found at redshift as high as @xmath03 ( @xcite ) , and may have contributed to the early pollution of the igm ( @xcite ) . the driving mechanism for the energy and metal pollution of the igm / icm is the feedback from star formation . in the basic scenario ,
the combined effects of ejecta and energy release from supernovae and massive stars in regions of intense star formation form superbubbles that sweep up the surrounding medium and produce cavities containing hot , shocked gas ( ( * ? ? ?
* weaver et al .
if the bubble has enough energy to expand to a size of a few times the scale height of the ambient medium , the swept - up shell may break out and the wind internal to the bubble may expand into the igm ( ( * ? ? ?
* chevalier & clegg 1985 ) ) .
many theoretical and observational studies have investigated the mechanical feedback from star formation and its efficiency in transfering energy and material to regions far removed from the site of star formation ( ( * ? ? ?
* e.g. , de young & heckman 1994 ) ; ( * ? ? ?
* suchkov et al . 1994 ) ; ( * ? ? ?
* mac low & ferrara 1999 ) ; ( * ? ? ?
* strickland & stevens 2000 ) ; ( * ? ? ?
* heckman , armus & miley 1990 ) ; ( * ? ? ?
* dahlem 1997 ) ; ( * ? ? ?
* ferrara & tolstoy 2000 ) ; ( * ? ? ?
* cecil , bland - hawthorn & veilleux 2002 ) ) , as well as in triggering renewed star formation ( ( * ? ? ?
* e.g. , wada & norman 2001 ) ) .
yet many questions remain to be fully answered . among these , progressing from small to large scales : the impact of the spatial distribution , history , and duration of the star forming event ( ( * ? ? ?
* clarke & oey 2002 ) ) ; the coupling of the feedback from star formation with the surrounding ism ; the dependence of the feedback efficiency on the conditions of the ism ( e.g. porosity ; ( * ? ? ?
* clarke & oey 2002 ) ) ; and the role of the global galaxy parameters ( mass , metallicity , environment ; ( * ? ? ?
* dahlem 1997 ) ; ( * ? ? ?
* strickland et al . 2004 ) ) .
a review of some of these issues is given by d. strickland ( 2004 , these proceedings ) .
one of the complicating factors in tackling the various facets of feedback is the vast range of physical scales that needs to be probed . while the relevant scales for charting star formation are those of the stars and star clusters ( a few pc ) , a study of the structure and conditions of the ism needs to sample scales of tens to hundreds of pc .
moving on to larger scales , superwinds are best investigated on kpc scales , while probing the influence of global parameters involves scales of tens of kpc ( galactic sizes ) to mpc ( interactions ) . even for the closest galaxies , the small to intermediate scales relevant for the interaction feedback
ism are accessible only via high angular resolution observations ( e.g. , with the hubble space telescope ) , as 10 pc=0.4@xmath1 for a galaxy at 5 mpc distance .
the study reviewed here involves four local starburst galaxies observed with the wide field planetary camera 2 on the hst . because of their intense - to - extreme star formation rates , starburst galaxies are the sites where stellar feedback can have its most dramatic influence on the structure and evolution of the ism ( and surrounding igm ; @xcite ) , and are , therefore , optimal laboratories to investigate mechanical feedback .
the four galaxies in the present study cover a range in luminosity , metallicity , star formation rate , and environment , but they are all closer than 5 mpc ( table [ tab1 ] )
. this investigation attempts at addressing the issue of the coupling feedback
ism , and its relation to the duration of the star forming event .
the problem is tackled from two fronts : stellar populations and ism conditions .
the ages of the young ( @xmath2300 myr ) stellar clusters are used to set a minimum value to the duration of the current starburst event , while measurements of the shocked gas provide a constraint on the fraction of starburst mechanical output that is recovered in the ism within a small distance ( @xmath21 kpc ) of the site of star formation .
.properties of the observed galaxies .
columns are as follows ; @xmath3 : distance ; @xmath4 : physical size of the region observed ; @xmath5 : absolute b magnitude ; @xmath6 : oxygen abundance , 12@xmath7log(o / h ) ; @xmath8 : star formation rate derived from the extinction corrected h@xmath9 luminosity ; @xmath10 : fraction of the h@xmath9 luminosity associated with shocks ; @xmath11 : ratio of the predicted - to - observed h@xmath9 luminosity associated to shocks .
ngc3077 is in the m81 group , in close interaction with m81 itself ; ngc4214 is an isolated galaxy , while ngc5236 and ngc5253 form a loose pair .
[ cols= " < , > , > , > , > , > , > , > " , ]
the star cluster populations and the spatial distributions of their ages within the starburst sites were investigated for three of the galaxies listed in table [ tab1 ] , ngc3077 , ngc5236 , and ngc5253 ( @xcite , @xcite ) . for ngc5253 , the diffuse starburst population had previously been age - dated ( @xcite , @xcite ) , which , combined with the more recent study of its cluster population , yields a complete picture of the recent star formation history in this galaxy .
images of the galaxies in the hst / wfpc2 medium and broad filters uv , v , i , and in the h@xmath9 and h@xmath12 line emission provide enough angular resolution to identify clusters , and enough color information to age - date them .
in particular , two age sensitive diagnostics , the color color diagram uv@xmath13v versus v@xmath13i and the equivalent width of h@xmath9 , were combined to constrain the most likely age for each stellar cluster and , assuming a salpeter imf , to assign a mass ( figure [ fig1 ] ; @xcite , @xcite ) .
h@xmath9/h@xmath12 ratio maps were used to correct photometry and fluxes for the effects of dust reddening ( @xcite ) .
the cluster ages were then re - mapped onto the galaxy images to derive the history of the cluster formation in each starburst .
the two dwarf galaxies , ngc3077 and ngc5253 , show time
extended star formation , spanning the last 100300 myr , and a very ` chaotic ' spatial distribution of the cluster ages .
this similarity between the two galaxies is counterbalanced by striking differences . in ngc3077 , there are star clusters as old as 300 myr , while in ngc5253 the oldest clusters unambiguously detected in the hst images do not seem to be significantly older than @xmath020 myr ( @xcite ) . indeed , the time
extended star formation history for ngc5253 is inferred not from its clusters , but from its diffuse stellar population , whose colors are consistent with constant star formation over the past @xmath0100 - 200 myr ( @xcite ) .
the dearth of ` old ' star clusters in ngc5253 could be caused by rapid cluster dissolution timescales in the center of the galaxy ; by rescaling the milky way model of @xcite , @xcite and @xcite derive dissolution timescales in the range 1650 myr for ngc5253 .
these short timescales are driven by the high central velocity dispersion measured in the galaxy ( @xcite ) .
a smaller velocity dispersion may be present in ngc3077 , as inferred from molecular cloud velocities ( @xcite ) , implying longer survival times for its clusters .
contrary to the dwarfs , the distribution of the stellar clusters in the center of the giant spiral ngc5236 shows a high level of organization , possibly a reflection of the different dynamical environment ( presence of a bar ) in the massive galaxy .
the uv bright clusters are distributed along a half ringlet surrounding the optically bright nucleus .
the distribution of the ages also shows a high level of correlation in the star formation , which appears propagating in the s
n direction and inside out in the ringlet ( ( * ? ? ?
* harris et al . 2001 ) , @xcite ) .
there is a clear peak in the number of clusters in the narrow age range 57 myr , with a handful of older clusters ( up to @xmath050 myr ) .
although the diffuse population is generally consistent with constant star formation over the past @xmath01 gyr , the sharp cut off of the cluster ages around 7 myr could still be explained with either a recent starburst or very rapid cluster dissolution timescales ( @xcite ) . independently of whether the current starburst in the grand design spiral is confirmed to be younger than @xmath010 myr , a clear mechanism for the triggering and feeding of the starburst , i.e. gas infall along the main bar , is present in ngc5236 ( @xcite ) .
no such obvious mechanism is present in the two dwarf galaxies .
their starbursts may have been triggered by past interactions with their companions ( m81 for ngc3077 , and ngc5236 for ngc5253 ) , but these interactions occurred a few hundred myr ago , and are unlikely to be the direct cause of the starbursts observed today .
the stellar population study of these galaxies was complemented with the investigation of the conditions of the ism within and around the four starbursts .
the standard emission line diagnostic diagram log([oiii]/h@xmath12)log([sii]/h@xmath9 ) was used to discriminate photoionized from non photoionized gas ( @xcite ) .
hst / wfpc2 images were obtained with narrow band filters in correspondence of the forbidden lines [ oiii](5007 ) and [ sii](6726 ) , and of the hydrogen recombination lines h@xmath9 and h@xmath12 .
the diagnostic diagram was constructed by dividing each image in a grid of bins 36 pc in size , measuring line fluxes in each bin 5 @xmath14 above the noise level , and deriving line ratios from spatially corresponding bins .
figure [ fig2 ] ( left panel ) gives an example of the resulting diagrams ( @xcite ) .
the ` maximum starburst line ' defined by @xcite was used as the separating line between photoionized gas ( to the bottom left of the line ) from non photoionized gas ( to the top
right of the line , figure [ fig2 ] ) .
this can be considered a conservative definition of non photoionized gas , yet it yielded clear identifications in all four galaxies .
the non photoionized gas in these galaxies shows two basic morphologies : enclosed bubbles and shells / filaments ( figure [ fig2 ] , right panel ) , and is likely due to shock ionization ( @xcite ) .
indeed , the presence of star - formation - induced shocks is supported by observations of diffuse x ray emission in the centers of all four galaxies . the shock
ionized component is responsible for a small fraction , about 3%4% , of the total h@xmath9 luminosity ( table [ tab1 ] , column 7 ) , but is spread over a significant fraction of the area covered by the h@xmath9 emission , about 15%25% . the availability of multi color ( uv to
i ) and h@xmath9 images allows us to place contraints on the characteristics of the ionizing stellar population , and , once coupled with models , on the expected amount of mechanical energy coming out of the starburst .
the h@xmath9 luminosity predicted from the mechanical starburst output can then be compared with the _ observed _ h@xmath9 luminosity associated with shocks .
the last column of table [ tab1 ] shows the predicted - to - observed h@xmath9 luminosities from shocks , with the range marking starbursts of different durations , from 10 myr ( left - hand - side figure ) to 100 myr ( right - hand side figure ) . in this column , a figure smaller then 1 means that the starburst does not produce enough mechanical energy to account for the observed h@xmath9 luminosity of shocks , while a figure larger than 1 means the starburst can support the observed shock luminosity . for the three dwarfs ,
between 70% and 100% of the mechanical energy is deposited within the immediate surroundings of the starburst , roughly within the inner 0.21 kpc .
furthermore , starbursts with durations shorter than @xmath030 myr do not produce enough mechanical energy to account for the observed h@xmath9 luminosity of shocks .
it is worth reminding that we adopted a conservative criterion for identifying shocked gas in the galaxies , thus the limitation on the minimum duration of the starbursts is quite stringent .
in addition , the result is independent of whether the parent galaxy is isolated or in interaction with other galaxies . since the vast majority of the starburst mechanical output is recovered in the immediate surroundings of the starburst itself , it is not unlikely that the shocks can trigger renewed star formation .
this inference could help explain the long star formation timescales required by the high luminosity of the h@xmath9 emission associated with shocks , and the even longer timescales derived from the stellar population studies .
for the giant spiral , only between 35% and 75% of the available mechanical output is recovered in the surrounding ism , for a variety of starburst conditions .
thus , there are no restrictions on the current starburst coming from the characteristics of the shocked gas .
in addition , the non photoionized gas is located in two main bubbles , and there is a striking absence of extended features ( shells or filaments ) .
this suggests a confined starburst .
the characteristics of the stellar populations and ism in the starbursts of four nearby galaxies suggest a strong coupling between the starbursts and the surrounding ism .
this is particularly true for the three dwarf galaxies in the sample . here
the amount of observed h@xmath9 luminosity associated with shocks corresponds to about 70%100% of the total mechanical output from the starbursts , and places a lower limit of about 30 myr to the duration of the starbursts ( i.e. , these can not be ` bursts ' ) .
such large fractions of mechanical energy available in the immediate surroundings of starbursts may suggest triggered star formation as a mechanism to explain the long duration timescales inferred from stellar population studies ( @xmath15100300 myr ) , in the absence of direct , recent triggers .
the author would like to thank the local organizing committee , and in particular thaisa storchi - bergmann , for their hospitality .
this work was supported by nasa long term space astrophysics grant nag-9173 and by nasa hst grant go-9144 . | the interplay between stellar populations and gas in local starburst galaxies is analyzed using images from the hubble space telescope to map the ages of the young stellar components and to isolate the contribution of shocks on spatial scales ranging from a few tens of pc to @xmath01 kpc .
the shocked gas represents a small fraction of the total ionized gas in these objects , yet it can have profound effects on the long term evolution of the starburst , which may include the triggering of new star formation . |
the @xmath6 kev extragalactic background light is dominated by sources with fluxes around @xmath7 @xcite . to obtain a more complete picture of the history of production in active galactic nuclei ( agns ) , these sources , which have surface densities of only a few hundred deg@xmath8 , need to be studied in more detail . since the chandra deep field ( cdf ) surveys ( cdf - n , @xcite and @xcite ; cdf - s , @xcite ) cover too little solid angle ( 0.12 and 0.11 deg@xmath9 , respectively ) to provide a large sample of such sources , we have undertaken a wide - area ( 0.4 deg@xmath9 ) , contiguous _ chandra _ survey of the lockman hole northwest field .
the properties of the sources in our survey called the _
large area synoptic x - ray survey , or clasxs are described in the companion paper by @xcite ; hereafter , y04 .
we designed clasxs to sample a large , contiguous solid angle , while remaining sensitive enough to measure sources @xmath10 times fainter than the observed break in the @xmath6 kev log @xmath11log @xmath12 distribution ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we chose a survey size of about 0.4 deg@xmath9 based on _ asca _ observations , which show a 6% rms variance for hard intensities on a scale of @xmath13 deg@xmath9 @xcite .
this variance is believed to be responsible for the discrepancies in the number counts among the ultradeep _ chandra _ surveys @xcite . with clasxs , we are able to examine the large scale structure of the sources @xcite , something that can not be done by combining random , non - contiguous _ chandra _ pointings .
clasxs is the essential step between the ultradeep , narrow _ chandra _ surveys and wide - field , shallow surveys , such as those obtained by _
rosat _ and _
asca_. in this paper , we present photometric and spectroscopic data of the optical counterparts to the clasxs sources .
we briefly describe the observations in 2 .
( a full discussion may be found in y04 . )
we discuss our photometric observations in 3 , our spectroscopic data in 4 , and the photometric and spectroscopic properties of the clasxs sources in 5 and 6 , respectively . in 7 , we present the colors and luminosities of the clasxs sources . in
8 and 9 , respectively , we discuss our results and summarize our main conclusions .
we use j2000 coordinates and the cosmological parameters @xmath14 km s@xmath15 mpc@xmath15 , @xmath16 , and @xmath17 .
to minimize the effects of galactic attenuation , we chose to position our clasxs survey in the lockman hole ( specifically , the lockman hole - northwest , one of two _ infrared space observatory _ [ _ iso _ ] fields observed by @xcite ) , which has the lowest integrated galactic column density in the sky ( n@xmath18 cm@xmath8 ; * ? ? ? * ) .
@xcite find an additional ionized hydrogen component that comprises @xmath20 of the attenuating material ; however , limitations on their data prevent a calculation of the precise ionization fraction .
regardless , the total hydrogen content is too low to significantly affect the properties . for reference
, we note that our field is located @xmath21 northwest of the _ rosat _ lockman hole field ( @xcite ) , which was also observed by _ asca _ ( @xcite ) , _ bepposax _ ( @xcite ) , _ chandra _ ( @xcite ) , and _ xmm - newton _ ( @xcite ) .
clasxs consists of nine contiguous _ chandra _ acis - i exposures approximately arranged in a @xmath22 grid centered at @xmath23 ( 10:34:02.1 , + 57:46:25.0 ) .
the aim points are separated by about @xmath24 .
the central field has a longer integration ( 70 ks ) than the surrounding eight 40 ks fields .
the total field coverage is @xmath25 deg@xmath9 , with 70 ks aim point flux limits ( y04 ) of @xmath26 ( soft ; @xmath27 kev ) and @xmath28 ( hard ; @xmath6 kev ) . we have detected 525 independent sources in at least one of the three bands . in y04 ,
the fluxes for the clasxs sources were calculated from the observed counts using individual power - law indicies that were based on the ratio of the hard - to - soft counts ( i.e. hardness ratios ) .
here we have converted the @xmath29 kev y04 fluxes into @xmath30 kev fluxes using a power - law spectral energy distribution with a photon index of @xmath31 and a column density of n@xmath32 cm@xmath8 .
if a source is not detected in all of the bands , then an upper limit for the flux in any undetected band is calculated by measuring the flux at the source position ( as measured in the detected band ) using a circular aperture and removing the local background signal .
if the local background flux is larger than the flux measured at the source position , then the background flux is used as an upper limit for the source .
additional information regarding the clasxs catalog , including reduction and analysis procedures , can be found in y04 .
all 525 clasxs sources from y04 are listed in table [ main_table ] in the appendix , which also gives the optical magnitudes and spectroscopic redshifts , when available ( see 3 and 4 ) .
we obtained deep broadband johnson _ b _ and _ v _ , cousins _ r _ and _ i _ , and sloan @xmath4 observations reaching @xmath33 ab magnitudes of @xmath34 , @xmath35 , @xmath36 , @xmath37 , and @xmath38 , respectively , with suprime - cam @xcite on the subaru 8.2 m telescope . using the canada - france - hawaii telescope ( cfht ) with the cfh12k camera , we obtained wider field images in broadband johnson _ b _ , cousins _ r _ , and cfht @xmath4 reaching @xmath33 ab magnitudes of @xmath39 , @xmath36 , and @xmath40 .
table [ data_details1 ] summarizes the telescopes , dates of the observations , and the integration times .
table [ data_details2 ] summarizes the average seeing , @xmath33 limits , and areas of each image .
the images and the optical source catalog are presented in @xcite , in which a more extensive discussion of the observations and reductions can be found .
we briefly summarize the suprime - cam and cfh12k observations and reductions in the appendix .
we performed source detection on the _ r_-band image using the program sextractor @xcite . to improve faint source detection and minimize sky noise ,
the image was smoothed using a @xmath41 gaussian kernel .
the detection threshold was set to be eight contiguous pixels with counts at least @xmath42 above the sky background .
we measured the optical magnitude of each source with the idl routine aper , using a 3 diameter aperture centered on the optical counterpart , if detected , or on the position if not .
the magnitudes of bright sources that are significantly larger than the 3 aperture are underestimated , as are @xmath43 sources , which are typically saturated in our images .
table [ main_table ] in the appendix gives the optical properties of the clasxs x - ray sources .
the clasxs source number ( from y04 ) is given in column ( 1 ) .
the positions for the detected optical counterparts are given in columns ( 5)-(6 ) .
the aperture - corrected @xmath0 , @xmath1 , @xmath2 , @xmath3 , and @xmath4 magnitudes are given in columns ( 9)-(13 ) .
column ( 14 ) gives the measured redshifts of identified sources . in this column ,
we labeled spectroscopically identified stars `` star '' . if we spectroscopically observed a source but did not identify it , then we labeled the source `` obs '' .
the aperture correction was calculated by examining the curve of growth for isolated , moderately bright ( @xmath44 ) sources .
the 3 aperture corrections for the subaru ( cfht ) images are @xmath45 ( @xmath46 ) , @xmath47 , @xmath48 ( @xmath49 ) , @xmath50 , and @xmath51 ( @xmath47 ) , for the _ b _ , _ v _ , _ r _ , _ i _ , and @xmath4 bands , respectively .
sources that fall outside of the field of view of a given filter are left blank . if a source falls within the field of view but is not detected , then it is assigned the @xmath42 magnitude limit for that filter and is considered an upper limit .
_ chandra _ s cylindrical mirrors were designed to be the smoothest mirrors ever constructed in an effort to obtain the highest resolution x - ray images ever ( @xmath52 at the aim point ) , and thus the most accurate positions for the sources .
this unprecedented positional accuracy allows the secure identification of counterparts at other wavelengths .
the cylindrical mirror design , which is required for the grazing incidence optics needed to focus , introduces an aberration that increases the point spread function ( psf ) of an image with an off - axis angle .
for example , _ chandra _ s psf at an off - axis angle of @xmath53 ( @xmath54 ) from the aim point is @xmath55 ( @xmath56 ) .
this increasing psf size with off - axis angle makes it difficult to identify optical counterparts using a fixed search radius .
if too small a radius is used , then counterparts for sources with large psfs may be missed .
if too large a radius is used , then it will negate _ chandra _ s excellent resolution and ability to securely identify sources with no detectable optical counterpart .
because of this , we divide the clasxs sample into two groups using the radial distance from the aim point ( @xmath57 ; col [ 4 ] in table [ main_table ] ) .
a 2 search radius is used for sources with @xmath58 , and 3 for sources with @xmath59 .
we use the smallest aim point distance if a source is detected in more than one of the clasxs pointings . since the nine individual clasxs fields that make up the larger mosaic seperated by only @xmath24 , the majority of our sources are less than @xmath60 from an aim point of one of these fields .
thus , all of our positions are relatively uniformly determined .
if more than one optical source falls within the search radius , then we designate the closest optical neighbor as the counterpart . using these criteria , 484 ( @xmath61 ) of the 525 detected sources in the y04 catalog have optical counterparts , and 264 ( @xmath62 ) of these have magnitudes that are bright enough for straightforward spectroscopic follow - up ( @xmath63 ) . in figure [ astrometry ]
, we show the ( @xmath64 optical ) astrometric offsets for the clasxs sources .
histograms for the right ascension and declination offsets are shown above and to the right , respectively .
the average astrometric discrepancies in both axes are @xmath65 arcseconds .
we ran simulations to examine the probability of an source being assigned an incorrect optical counterpart .
the probability of a chance projection is a strong function of the limiting optical magnitude in the catalog , since there are many more sources at the faint end .
the concentric circles represent the probability of a source with a random right ascension and declination being assigned an optical counterpart from the full optical catalog ( @xmath66 ) .
the inner circle is 10% , and the other circles increase outward with 10% increments .
the majority of clasxs sources have ( @xmath64 optical ) separations of less than @xmath67 ( @xmath68 ) for which there is only a 2% probability of an incorrect /optical match . with our search radius of @xmath41
, we calculate that 30% of the clasxs sources that in fact have no optical counterpart will be assigned an incorrect optical counterpart .
almost all of these will be with optically faint sources .
if we limit the optical sources to those spectroscopically accessible ( @xmath63 ) , we find only an 8% chance of an incorrect /optical overlap using a @xmath41 matching radius .
we show @xmath2-band thumbnails for 521 of the 525 clasxs sources in the y04 catalog in figure [ thumbnails ] in the appendix , ordered by increasing right ascension .
four sources ( 192 , 318 , 328 , and 359 ) are outside the fields of view of our optical images and are not included .
stars are labeled `` star '' .
optically undetected sources are labeled `` b '' for `` blank '' . spectroscopically observed sources that could not be identified are labeled `` obs '' .
optical spectra were obtained using the multi - fiber spectrograph hydra @xcite on the wiyn 3.5 m telescope for bright ( @xmath69 ) sources , and with the deep extragalactic imaging multi - object spectrograph ( deimos ; * ? ? ?
* ) on the 10 m keck ii telescope for fainter sources .
table [ spec_obs_summary ] summarizes the telescopes , instruments , and dates of the spectroscopic observations .
@xmath70 kev flux distribution for the 525 clasxs sources ( _ shaded _ , spectroscopic redshifts ; _ hatched _ , spectroscopically observed sources that could not be identified ; _ open _ , spectroscopically unobserved sources ) .
the arrow marks the flux where the source contribution to the @xmath6 kev xrb peaks . ] for the hydra observations , we used a low - resolution grating , [email protected] , centered at 7600 , yielding a wavelength coverage of @xmath71 with a resolution of 2.64 pixel@xmath15 .
the red bench camera was used with the 2 `` red '' hydra fibers to maximize the sensitivity at longer wavelengths . to obtain a wavelength solution for each fiber , cuar comparison lamps
were observed in each hydra configuration .
our hydra masks were designed to maximize the number of optically bright sources in each configuration , while minimizing the amount of overlap between configurations .
fibers that were unable to be placed on a source were assigned to a random sky location .
we observed 2.7 hrs on two hydra configurations in 2001 february , 7.4 hrs on two configurations in 2002 february , and 6.0 hrs on two configurations in 2002 march . to remove fiber - to - fiber variations ,
on - source observations were alternated with @xmath72 `` sky '' exposures taken with the same exposure times .
this effectively reduced our on - source integration times to 1.3 , 3.6 , and 3.0 hrs in 2001 february , 2002 february , and 2002 march , respectively .
reductions were performed using the standard iraf package dohydra . to optimize sky subtraction
, we performed a two - step process . in the first step ,
the dohydra routine was used to create an average sky spectrum using the fibers assigned to random sky locations .
this average sky spectrum was then removed from all of the remaining fibers . in the offset images ,
this step effectively removed all of the sky signal , leaving behind only residuals caused by differences among the fibers . to remove these variations
, we then subtracted the residuals present in the sky - subtracted offsets from the sky - subtracted , on - source spectra .
we found this method to be very effective at removing the residuals created by fiber - to - fiber variations with hydra .
for the deimos observations , we used the 600 lines mm@xmath15 grating , which yielded a resolution of 3.5 and a wavelength coverage of 5300 .
the exact central wavelength depends upon the slit position , but the average was 7200 .
each @xmath73 hr exposure was broken into 3 subsets . in each subset
the object was stepped 1.5 in each direction .
the deimos spectroscopic reductions follow the same procedures used by @xcite for lris reductions .
the sky contribution was removed by subtracting the median of the dithered images .
cosmic rays were removed by registering the images and using a cosmic ray rejection filter on the combined images .
geometric distortions were also removed and a profile - weighted extraction was applied to obtain the spectrum .
wavelength calibration was done using a polynomial fit to known sky lines rather than using calibration lamps .
the spectra were individually inspected and a redshift was measured only for sources where a robust identification was possible .
the high - resolution deimos spectra can resolve the doublet structure of the [ ] @xmath74 line , allowing spectra to be identified by this doublet alone .
figure [ log_flux_hist ] shows the @xmath6 kev flux distribution for the clasxs sources .
the sources that were spectroscopically observed _ ( hatched ) _ and spectroscopically identified _
( solid ) _ are indicated .
the arrow marks the flux where the source contribution to the @xmath6 kev xrb peaks .
of the 525 clasxs sources detected in y04 , we have spectroscopically observed 467 ( 90% ) and identified 271 ( 52% ) , including 20 stars .
tables [ optical_soft ] and [ optical_hard ] give the median optical magnitudes for the clasxs soft and hard sources , respectively .
the tables also give , in parentheses , the median optical magnitudes for the clasxs sources combined with other observations from _ chandra _ @xcite , _ rosat _
@xcite , _ xmm - newton _
@xcite , and _ asca _ @xcite .
figure [ r_mag_vs_sb ] shows the _ r _ magnitudes of the combined samples versus ( a ) soft and ( b ) hard flux .
any of the published catalogs that used energy ranges for the soft and hard bands other than @xmath75 kev and @xmath6 kev , respectively , were converted to this range using the power law assumed by the respective authors , or , if none were given , then by assuming a power law of @xmath31 and no absorption .
the clasxs sources with optical upper limits are denoted by arrows .
clasxs sources brighter than @xmath76 are typically saturated in our images , so the measured magnitudes of these sources may be underestimated .
the shaded regions in figures [ r_mag_vs_sb]_a _ and [ r_mag_vs_sb]_b _ represent typical ranges of log@xmath77 for different source classes ( e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
the flux in the _ r_-band , @xmath78 , is related to the _ r _ magnitude by log@xmath79 @xcite .
agns typically lie in or above the region defined by the loci log@xmath80 _ ( lightest shading ) _ , while normal galaxies lie below log@xmath81 _ ( darkest shading)_. note that the flux limit of the clasxs survey ( @xmath82 ) makes it an agn dominated survey .
the median @xmath2-band magnitudes in each flux bin ( see values in parentheses in tables 4 and 5 ) are shown in figure [ r_mag_vs_sb ] as the large diamonds with @xmath83 standard deviations for each flux bin , denoted by error bars .
the median /optical flux ratios for the soft sources follow the log@xmath84 loci for fluxes larger than @xmath85 , but drop to smaller /optical ratios at fainter fluxes .
as noted by @xcite , the explanation for this is straightforward . at bright soft fluxes ,
the agns usually dominate the and optical light , so the galaxies follow a constant /optical flux ratio . at fainter fluxes , the starlight begins to dominate over the optical agn contributions , possibly as a result of obscuration , an increase in the number of intrinsically lower luminosity agns , or , at the faintest fluxes , an increase in the number of galaxies without agns ( e.g. * ? ? ?
since stars contribute less than the agns to the emission , the result is a drop in the /optical flux ratios .
in addition , any intrinsic obscuration will also attenuate the soft emission from the central engine . for near compton - thick agns ,
all of the soft and optical agn light will be attenuated .
the @xmath2-magnitudes shown in figure [ r_mag_vs_sb ] are not nuclear magnitudes , so they differ from the optical/ studies done of nearby seyfert galaxies .
we use a fixed aperture for our photometric measurements that covers a varying fraction of the optical counterpart .
this fraction is dependent on the apparent angular size of the optical source and can vary between the nuclear magnitude for nearby sources and the total magnitude for high - redshift sources .
this introduces a scatter in the y - axis of figure [ r_mag_vs_sb ] , because a high - redshift source will have a brighter optical magnitude than its low - redshift twin .
since most of our sources have small angular sizes ( see figure [ thumbnails ] ) , our @xmath2-band magnitudes can be considered upper limits to the nuclear @xmath2-band magnitudes . in other words , if we could measure only the nuclear magnitudes for the clasxs sources , then the observed @xmath2-band magnitudes would decrease .
this would shift many clasxs sources away from the quiescent galaxies regime and into the agn regime .
the presence of an obscuring torus can affect the /optical flux ratio of a source , depending on which band we examine . since a photon s cross - section decreases rapidly with increasing photon energy , light from the near - infrared to soft is typically attenuated in moderately obscured agns , while hard light penetrates all but the most obscured sources .
figure [ fr_vs_hr ] shows the @xmath75 kev to @xmath2-band and @xmath6 kev to @xmath2-band flux ratios versus soft - to - hard flux ratio for the clasxs soft and hard selected samples .
the samples were further separated by hard flux into bright ( @xmath86 ) and faint ( @xmath87 ) categories .
the filled squares denote broad - line agns ( blagns ) .
the large diamonds are the average @xmath88 within the soft - to - hard flux ratio ranges of @xmath89 , @xmath90 , @xmath91 , @xmath92 , and @xmath93 .
we expect that for the @xmath75 kev /optical flux ratio , the presence of any obscuring material will attenuate both the soft and optical photons , making the flux ratio only weakly dependent on the column density for unobscured to moderately obscured sources .
moreover , for highly obscured sources , the agn contribution to both the observed soft and optical light should become negligible , causing the galaxy to descend into the `` quiescent galaxy '' category .
if we treat the soft - to - hard flux ratio as an approximation to the line - of - sight column density ( column density would decrease with increasing soft - to - hard flux ratio ) , we can see from figure [ fr_vs_hr ] that the soft /optical flux ratio _ ( top row ) _ does indeed remain relatively constant for @xmath94 .
moreover , at larger column densities , where both the optical and the soft light are obscured , there may be a slight drop in the soft /optical flux ratio as the agn contributions to both become negligible and the optical starlight from the galaxy begins to dominate .
in contrast , we expect that moderate column densities of material will not significantly attenuate the hard photons from an agn , so moderately obscured sources will have high hard /optical flux ratios . from figure [ r_mag_vs_sb ] _
( bottom row ) _ , we see that as the column density increases , the agn s optical light is blocked , while the hard can still penetrate the material .
thus , the hard /optical flux ratio increases until the material becomes opaque to hard . at these high column densities ,
even the light is absorbed , and we would expect to see the /optical flux ratio begin to drop
. however , near - compton - thick sources are very weak and difficult to detect , and there may be very few of them in the sample .
the aforementioned trends in the soft and hard -to - optical ratios appear to hold for both the bright and faint clasxs sources
. however , there are noticeable differences in the two populations that can be seen in figure [ fr_vs_hr ] .
there is a significantly larger population of blagns in the bright population than in the faint population .
this is consistent with the discovery by @xcite that the blagns dominate the number densities at the higher luminosities , while the non - blagns dominate at the lower luminosities .
the redshift distribution of the clasxs sources is shown in figure [ redshift_hist ] .
the two histograms have @xmath95 and @xmath96 width bins
. there are no obvious structures at @xmath5 in the clasxs redshift distribution , unlike the apparent excesses found at @xmath97 and @xmath98 in the cdf - n sample by @xcite , and at @xmath99 and @xmath100 in the cdf - s sample by @xcite and @xcite .
( there might be a grouping of clasxs sources at @xmath101 , though this is not easily seen in the figure . ) the differences between the redshift distributions of clasxs and the cdfs show that we have indeed sampled a large enough area in clasxs to average out the large scale structure .
redshift distribution for the spectroscopically identified clasxs sources .
all stars and sources without measured redshifts have been excluded .
two different bin sizes ( @xmath95 and @xmath96 ) are shown . ] following @xcite , we optically classified sources into four categories , roughly using the procedure of @xcite to analyze low - redshift radio samples .
we classified sources without any strong emission lines ( ew([])@xmath102 or ew(h@xmath103 @xmath104 ) as absorbers ; sources with strong balmer lines and no broad or high - ionization lines as star formers ; sources with [ ] or lines or strong [ ] ( ew ( [ ] 5007 ) @xmath105 ew(h@xmath106 ) ) as seyfert galaxies ; and , finally , sources with lines with fwhm @xmath107 km s@xmath15 as blagns . table [ source_type ] breaks down the identified sources by spectral type .
redshift vs. _ r_-band magnitude for the clasxs hard sample ( _ stars _ , stars placed at @xmath108 ; _ plus signs_,spectroscopically observed but unidentified sources placed at @xmath109 ; _ crosses_,spectroscopically unobserved sources placed at @xmath110 ; _ filled squares_,blagns ; _ open squares_non - blagns ) .
the graction of the clasxs sources that have been spectroscopically observed ( identified ) for each bin of @xmath111 mag ( _ dotted lines _ ) is shown across the top of the figure . ]
figure [ r_mag_vs_redshift ] shows redshift versus @xmath2-band magnitude for the clasxs hard sample , with blagns denoted by solid squares and non - blagns denoted by open squares .
figure [ redshift_hist_type ] combines the clasxs data with the data of @xcite , @xcite , and @xcite , dividing the sources into blagns and non - blagns . to remove the bias introduced by the depth of the cdf - n survey , which is deep enough to detect the emission from nearby galaxies with no agn component
, we have only shown the sources with @xmath112 .
the combined surveys have very few blagns at @xmath113 , but the numbers increase at @xmath114 and appear to decrease only slightly over the range @xmath115 before falling off beyond .
the redshift distribution is quite different for the non - blagn population .
however , there are strong selection effects in the spectroscopic identifications .
there are many more agns at low redshifts that do not have broad optical emission lines , either because of low intrinsic agn luminosity or obscuration .
thus , the identifications for these sources are based on their galaxy spectra , and at high redshifts , galaxy spectra are much harder to spectroscopically identify than blagn spectra , particularly in the spectroscopic redshift desert at @xmath116 ( e.g. , * ? ? ?
* ; * ? ? ?
redshift distribution of ( _ a _ ) blagns and ( _ b _ ) non - blagns with @xmath117 ( see 6 ) in four surveys : clasxs , cdf - n ( barger et al .
2003 ) , cdf - s ( szokoly et al .
2004 ) , and _ xmm - newton _ lockman hole ( mainieri et al .
all stars and sources without measured redshifts have been excluded . ]
in figures [ x - ray_flux_vs_b - r]_a _ and [ x - ray_flux_vs_b - r]_b _ , we plot the observed _
b@xmath64r _ color versus soft and hard flux for the spectroscopically identified clasxs sources with @xmath118 , excluding any sources that are saturated . we break the sources into three spectroscopic categories : blagns , seyfert galaxies , and `` normal '' galaxies ( the last category combines the absorbers and star formers classes of 6 ) .
the blagns are , on average , much bluer than the other types of sources .
this is due to the excess blue continuum from the agn itself that is commonly observed in unobscured ( type i ) agns .
thus , the colors and the spectroscopic classes are in good agreement .
distribution of absolute rest - frame @xmath0 magnitude for all of the clasxs sources with measured redshifts less than @xmath119 _
( open histogram)_. blagns ( non - blagns ) are shown as a solid ( _ hatched _ ) histogram .
( see 7 for an explanation of the redshift limit . ) ] the distribution of the absolute rest - frame _ b _ magnitudes for the clasxs sources is shown in figure [ abs_b_dist ] , calculated using the measured redshift and interpolating the _ b_-magnitude in the observer s rest - frame from the observed broadband colors .
since we do not have near - infrared observations of all of our sources , we can only use this technique to measure absolute _
b_-magnitudes for sources with @xmath120 . beyond this redshift ,
the rest - frame _ b_-band moves redward beyond our @xmath4 observations . only non - stellar sources with measured magnitudes in at least two wavebands are included .
blagns and non - blagns are shown as solid and hatched histograms , respectively .
our photometric data do not have the resolution necessary to separate the luminosity of the host galaxy from that of the central agn , so we can only comment on the combined agn+host luminosity .
since blagns have an excess blue continuum , one would expect a larger number of blagns at brighter absolute _ b _ magnitudes .
indeed , figure [ abs_b_dist ] shows that this is the case .
in figures [ redshift_flux]_a _ and [ redshift_flux]_b _ , we plot redshift versus soft and hard flux for the combined samples of 5 .
the data and symbols are identical to those in figure [ r_mag_vs_sb ] .
the clasxs data nicely fill in the intermediate flux interval between the deep _ chandra _ data and the shallower _ asca _ and _ rosat _ data .
the dashed lines show the median redshifts of the combined samples versus soft and hard flux .
the median redshifts remain around @xmath121 and do not appear to be strongly correlated with either soft or hard flux .
of course , the redshift determinations depend on the optical magnitudes of the sources .
since the magnitudes of high - redshift sources may be faint , these surveys might be biased against finding high - redshift sources .
the solid lines denote constant rest - frame luminosities , which were calculated assuming a @xmath31 power - law spectral energy distribution with no absorption .
they range from @xmath122 to @xmath123 ergs s@xmath15 ; the two largest luminosities are labeled .
any source more luminous than @xmath124 ergs s@xmath15 is very likely to be an agn on energetic grounds @xcite , though many of the intermediate luminosity sources do not show obvious agn signatures in their optical spectra ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we use the hard fluxes and redshifts of the combined samples in figure [ redshift_flux]_b _ to calculate the rest - frame hard luminosities . as in 2 , we assume a power law spectral energy distribution with @xmath31 for calculating the _ k_-corrections to determine the rest - frame hard luminosities . using a universal power law index rather than individual indices to calculate the _
k_-corrections results in only a small difference in the calculated luminosities @xcite .
figure [ lum_z ] shows the rest - frame hard luminosities for the hard sources and limiting fluxes of the clasxs , @xcite , @xcite , @xcite , and @xcite catalogs .
the clasxs sample has the largest number of @xmath125 sources .
even with the increased solid angle covered by the five surveys , there is an obvious lack of extremely luminous ( @xmath126 ) agns at @xmath113 .
the bolometric luminosity , @xmath127 , of an agn is related to the rate of accretion , @xmath128 , onto its supermassive black hole using the expression @xmath129 , where @xmath130 is the speed of light and @xmath131 is the accretion efficiency ( typically , @xmath132 ) . because agns are luminous from radio to energies , multiwavelength observations are needed to accurately calculate their bolometric luminosities .
if we use a bolometric correction of 35 @xcite appropriate for the type i agns to translate hard luminosity , @xmath133 , to @xmath127 and assume an efficiency of @xmath134 , then we can roughly estimate the accretion rates .
the results are shown in figure 11 _ ( right - hand axis)_. the small solid angle of the cdf - n survey prevented @xcite from determining if the lack of low redshift , high luminosity sources in their sample were due to an actual deficiency of these sources at low redshifts or simply a selection effect owing to the small local volume sampled and the low spatial density of these sources .
clasxs samples a larger local cosmological volume to better constrain the accretion rate at low redshifts , and we can now clearly see that the deficiency of these sources is indeed real and that at @xmath113 there are very few sources with accretion rates above a solar mass per year .
we have presented optical photometric and spectroscopic observations of the sources in the clasxs survey .
we obtained @xmath135 photometry for 521 of the 525 sources in the sample .
we spectroscopically observed 467 of the clasxs sources , obtaining redshift identifications for 271 .
the positions of the sources in our optical catalog agree well with the catalog , with average astrometric offsets of @xmath65 arcseconds in both right ascension and declination . using our optical data ,
we have examined the relationship between the and optical source properties .
we find that the average soft to optical flux ratio remains relatively constant for clasxs sources with low to moderate column densities of obscuring material .
conversely , the hard to optical flux ratio for these sources appears to be correlated with line - of - sight column density .
because of the large solid angle covered by clasxs , we do not find the low - redshift structures that are seen in narrower , deeper _ chandra _ surveys .
however , we do detect a possible excess of sources at @xmath101 . using hard catalogs from five _ chandra _ and _ xmm - newton _
surveys we find that there is a distinct lack of luminous , high - accretion rate sources at low redshifts ( @xmath136 , consistent with previous observations which showed that supermassive black hole growth is dominated at low redshifts by sources with low accretion rates @xcite .
we thank the referee , david alexander , for helpful comments that improved the manuscript .
we gratefully acknowledge support from nasa s national space grant college and fellowship program and the wisconsin space grant consortium ( a. t. s. ) , cxc grants go2 - 3191a ( a. j. b. ) and go2 - 3187b ( l. l. c. ) , nsf grants ast-0084847 and ast-0239425 ( a. j. b. ) and ast-0084816 ( l. l. c. ) , the university of wisconsin research committee with funds granted by the wisconsin alumni research foundation ( a. j. b. ) , the alfred p. sloan foundation ( a. j. b. ) , the david and lucile packard foundation ( a. j. b. ) , and the ids program of r. f. m.
details of the lockman hole - northwest photometric observations and reductions can be found in @xcite .
we provide a short summary of the subaru and cfht broadband photometric observations here . for the subaru images ,
bad pixels and chip gaps were filled in by creating mosaics using a five point star - shaped dither pattern with @xmath137 steps .
the separation between dithered images was made large enough to create deep sky flats during the image reduction process .
the camera was rotated by @xmath138 between dithers to remove bleeding from bright stars and provide better photometric calibration .
the suprime - cam data cover the central @xmath139 of the field , but only the central @xmath140 of the suprime - cam field goes to the full depth due to vignetting in suprime - cam . in the _ r _ band , we added four suprime - cam pointings covering @xmath141 from the subaru archive .
these data were collected using the same dither pattern and reduced by us in the same manner as our other data . for the cfht observations , a standard seven point
circular dither pattern was used .
the radius of the dither circle was @xmath142 , which was enough to fill the chip gaps but minimized the chip overlap .
the observing time was divided between two pointings separated by @xmath143 in the north - south direction .
this allowed us to cover a @xmath144 field while going deeper in a central strip .
all the data were reduced using nick kaiser s imcat tools .
flat - fielding of the cfht images was done using elixir .
the @xmath4 data were corrected for fringing .
the images were aligned astrometrically using a self - consistent astrometric solution described in @xcite .
the solution was calculated using the cfht @xmath2-band data along with wide - field suprime - cam data .
the usno - b1.0 @xcite was used as the reference catalog for this process and provided the absolute astrometric zeropoint .
all other data were warped onto this @xmath2-band image .
the individual mosaics were then combined .
thumbnails of 521 sources are shown in figure [ thumbnails ] . in table
[ main_table ] we present optical magnitudes and spectroscopic measurements , where available , for the clasxs point source catalog of y04 .
the 525 sources are ordered by increasing right ascension and are labeled with the y04 source number ( col .
the right ascension and declination coordinates of the sources are given in columns ( 2 ) and ( 3 ) , and the smallest off - axis angle is given in column ( 4 ) .
the right ascension and declination coordinates of the optical counterparts are given in decimal degrees in columns ( 5 ) and ( 6 ) .
the soft ( @xmath75 kev ) and hard ( @xmath6 kev ) fluxes , abbreviated sb and hb , are given in columns ( 7 ) and ( 8) .
the aperture corrected , broadband _
b , v , r , i , and @xmath4 _ apparent magnitudes are given in columns ( 9)-(13 ) .
the aperture correction was calculated by examining the curve - of - growth for isolated , moderately bright ( @xmath145 ) sources .
the 3 aperture corrections for the subaru ( cfht ) images are @xmath45 ( @xmath46 ) , @xmath47 , @xmath48 ( @xmath49 ) , @xmath50 , @xmath51 ( @xmath47 ) , for the _ b _ , _ v _ , _ r _ , _ i _ , and @xmath4 bands , respectively .
sources that fall outside of the field - of - view of a given filter are left blank . if a source falls within the field - of - view , but is not detected , then it is assigned the @xmath33 magnitude limit for that filter and is considered an upper limit .
the spectroscopic redshifts are given in column ( 14 ) .
lcccccrrrrrrrc 1 & 157.731750 & 57.55556 & 9.99 & 157.730803 & 57.55546 & 3.42 & 3.70 & & & 23.1 & & & + 2 & 157.748667 & 57.64561 & 8.76 & 157.747566 & 57.64558 & 1.62 & 5.30 & & & 22.7 & & & + 3 & 157.763833 & 57.61400 & 8.16 & 157.763506 & 57.61356 & 2.88 & 13.00 & & & 23.5 & & & obs + 4 & 157.775250 & 57.63011 & 7.80 & 157.774430 & 57.63017 & 3.33 & 0.74 & & & 23.8 & & 23.4 & obs + 5 & 157.799292 & 57.58936 & 7.27 & 157.799617 & 57.58951 & 2.88 & 0.91 & 24.7 & & 24.0 & & 23.0 & obs + @xmath0 & subaru 8.2 m & 2001 february 27 - 28 & 1.7 + @xmath0 & cfht 3.6 m & 2002 december 4 - 5 , 27 - 29 & 5.8 + @xmath1 & subaru 8.2 m & 2002 april 22 - 23 & 6.4 + @xmath2 & subaru 8.2 m & 2001 february 27 - 28 & 5.2 + @xmath2 & subaru 8.2 m & 2001 march 19 - 20 & 2.0 + @xmath2 & cfht 3.6 m & 2003 january 2 - 3 , 25 - 30 & 11.9 + @xmath3 & subaru 8.2 m & 2001 april 21 - 22 & 0.9 + @xmath146 & subaru 8.2 m & 2001 april 21 , 23 & 1.3 + @xmath147 & cfht 3.6 m & 2002 december 27 & 11.9 + @xmath147 & cfht 3.6 m & 2003 january 25 - 31 & 11.9 @xmath0 & subaru 8.2
m & 0.96 & 27.8 & 0.27 & 0.20 + @xmath0 & cfht 3.6 m & 0.97 & 27.6 & 0.49 & 0.49 + @xmath1 & subaru 8.2 m & 1.15 & 27.5 & 0.36 & 0.20 + @xmath2 & subaru 8.2 m & 0.96 & 27.9 & 0.27 & 0.20 + @xmath2 & subaru 8.2 m & 0.61 & 27.7 & 0.81 & 0.81 + @xmath2 & cfht 3.6 m & 0.89 & 27.9 & 0.49 & 0.49 + @xmath3 & subaru 8.2 m & 1.30 & 26.4 & 0.36 & 0.20 + @xmath4 & subaru 8.2 m & 1.01 & 26.2 & 0.36 & 0.20 + @xmath147 & cfht 3.6 m & 0.95 & 26.3 & 0.49 & 0.49 lll wiyn 3.5 m & hydra & 2001 february 19 - 20 + wiyn 3.5 m & hydra & 2002 february 6 - 7 + wiyn 3.5 m & hydra & 2002 march 9 - 10 + keck ii & deimos & 2003 january 28 - 29 + keck ii & deimos & 2003 march 26 + keck ii & deimos & 2003 april 24 - 26 + keck ii & deimos & 2003 may 26 + keck ii & deimos & 2004 january 17 - 19 + keck ii & deimos & 2004 april 13 - 15 lrcc 1 - 3 & 36 ( 248 ) & @xmath148 ( @xmath149 ) & 24.7 ( 23.3 ) + 3 - 10 & 149 ( 325 ) & @xmath150 ( @xmath151 ) & 24.6 ( 23.9 ) + 10 - 30 & 190 ( 349 ) & @xmath152 ( @xmath152 ) & 23.4 ( 23.2 ) + 30 - 100 & 92 ( 188 ) & @xmath153 ( @xmath153 ) & 22.4 ( 22.2 ) + 100 - 300 & 37 ( 80 ) & @xmath154 ( @xmath155 ) & 20.6 ( 20.7 ) + 300 - 1000 & 6 ( 31 ) & @xmath156 ( @xmath157 ) & 20.6 ( 18.6 ) 1 - 3 & 6 ( 125 ) & @xmath158 ( @xmath159 ) & 24.4 ( 22.7 ) + 3 - 10 & 47 ( 258 ) & @xmath160 ( @xmath151 ) & 23.7 ( 23.7 ) + 10 - 30 & 98 ( 356 ) & @xmath161 ( @xmath152 ) & 24.1 ( 23.8 ) + 30 - 100 & 197 ( 369 ) & @xmath162 ( @xmath163 ) & 24.0 ( 23.5 ) + 100 - 300 & 113 ( 236 ) & @xmath155 ( @xmath155 ) & 22.5 ( 22.3 ) + 300 - 1000 & 18 ( 64 ) & @xmath164 ( @xmath164 ) & 21.2 ( 20.7 ) + 1000 - 3000 & 3 ( 70 ) & @xmath165 ( @xmath166 ) & 20.7 ( 18.0 ) | we present photometric and spectroscopic observations of the sources detected in the wide - area , moderately deep _
chandra _ large area synoptic survey of the lockman hole - northwest field .
we have @xmath0 , @xmath1 , @xmath2 , @xmath3 , and @xmath4 photometry for 521 ( 99% ) of the 525 sources in the catalog and spectroscopic redshifts for 271 ( 52% ) , including 20 stars .
we do not find evidence for redshift groupings of the sources , like those found in the chandra deep field surveys , because of the larger solid angle covered by this survey .
we separate the sources by optical spectral type and examine the colors , apparent and absolute magnitudes , and redshift distributions for the broad - line and non - broad - line active galactic nuclei . combining our wide - area survey with other _
chandra _ and _ xmm - newton _ hard surveys , we find a definite lack of luminous , high accretion rate sources at @xmath5 , consistent with previous observations that showed that supermassive black hole growth is dominated at low redshifts by sources with low accretion rates . |
coronal mass ejections ( cmes ) are the most energetic eruptions in the solar system and can affect critical technological systems in space and on the ground when they interact with the geo - magnetosphere in high speeds ranging from several hundred to even more than one thousand kilometers per second .
therefore , it is an important issue to investigate the acceleration mechanisms of cmes in solar / space physics .
it is generally accepted that cmes are driven by magnetic flux rope ( mfr ) eruptions ( chen 2011 ) .
however , we can not observe mfr structures directly in the corona because no instrument can provide the high quality measurement of the coronal magnetic filed at present .
several lines of observations in the lower corona have been proposed as the proxies of mfrs , e.g. , filaments / prominences ( rust & kumar 1994 ) , coronal cavities ( wang & stenborg 2010 ) , sigmoids ( titov & dmoulin 1999 ; mckenzie & canfield 2008 ) , and hot channels ( zhang et al . 2012 ; song et al .
2014a , 2014b , 2015 ) . resolving the dynamics of these structures are critical to our understanding of the cme acceleration process .
* there exist two important magnetic energy release mechanisms : one is the resistive magnetic reconnection process ( carmichael 1964 ; sturrock 1966 ; hirayama 1974 ; kopp & pneuman 1976 ) and the other is the ideal global magnetohydrodynamic ( mhd ) mfr instability ( van tend & kuperus 1978 ; priest & forbes 1990 ; forbes & isenberg 1991 ; isenberg et al . 1993 ; forbes & priest 1995 ; hu et al . 2003 ; kliem & t " or " ok 2006 ; fan & gibson 2007 ; chen et al .
2007a , 2007b ; olmedo & zhang 2010 ) .
both mechanisms are supported by observations .
for example , good correlations exist between cme speed ( acceleration ) and the soft x - ray ( hard x - ray and microwave ) profiles of associated flares ( zhang et al . 2001 ; qiu et al .
2004 ; marii et al .
in addition , studies showed that the extrapolated magnetic flux in the flaring region was comparable with the magnetic flux of the mfr reconstructed from in situ data ( qiu et al .
all these cme - flare association studies support that reconnections play an important role in accelerating cmes . on the other hand
, statistical studies showed that the projected speed in the sky plane and kinetic energy of cmes only had weak correlations with the peak values of their associated x - ray flares ( yashiro et al .
2002 ; vrnak et al .
these observations support that the ideal mhd instability also makes significant contributions to the cme acceleration besides the magnetic reconnection . * * the theoretical and simulation studies also support that the instability and reconnection can accelerate the cmes ( dmoulin & aulanier 2010 ; roussev et al .
2012 ; amari et al .
generally , the instability plays an important role in triggering and accelerating the mfr , and then the magnetic reconnection accelerates the mfr further and allows the process to develop continuously ( priest & forbes 2002 ; lin et al .
* * however , it remains elusive whether both mechanisms have a comparable contribution to the acceleration in a particular event as they usually accelerate cmes in a close time sequence . in this letter
, we address this issue through analyzing a filament eruption with two apparently separated fast acceleration phases , instead of one as usual . *
the relevant observations and results are described in section 2 .
section 3 presents the related discussion , which is followed by a summary in section 4 .
the eruption process was recorded by the atmospheric imaging assembly ( aia ) telescope ( lemen et al .
2012 ) on board the _ solar dynamic observatory ( sdo)_. aia has 10 narrow uv and euv passbands with a cadence of 12 s , a spatial resolution of 1.2@xmath0 , and an fov of 1.3 @xmath1 .
the aia images shown in this letter ( figure 1 and supplementary movie ) are a small portion of the original full size images . the soft and hard x - ray ( sxr and hxr ) data shown in figure 3 are from _ geostationary operational environment satellite ( goes ) _ and the _ reuven ramaty high energy solar spectroscopic imager ( rhessi : _ lin et al .
2002 ) , respectively .
_ goes _ provides the integrated full - disk sxr emission from the sun , which are used to characterize the magnitude , onset time , and peak time of solar flares .
_ rhessi _ provides the hxr spectrum and imaging of solar flares .
the filament eruption originated from noaa active region 12151 located at the heliographic coordinates @xmath2s09e76 on 2014 august 24 .
this eruption produced an m5.9 class sxr flare on _ goes _ scale , which started at 12:00 ut and peaked at 12:17 ut .
a cme ( linear velocity 417 km s@xmath3 ) associated with it was recorded by the large angle spectroscopic coronagraph ( lasco , brueckner et al .
1995 ) on board the _ solar heliospheric observatory_. we inspect the aia images and find that the filament appeared in all bandpasses corresponding to the coronal and chromospheric temperatures , which indicate that it has a multi - thermal nature .
the eruption process snapshots observed with 304 ( @xmath20.05 mk ) , 171 ( @xmath20.6 mk ) and 335 ( @xmath22.5 mk ) are presented at the top , middle and bottom panels of figure 1 , respectively .
the time when the image was taken is shown at the top of each panel .
the arrows depict the euv brightening positions , where the plasma was heated by magnetic reconnection .
the full aia image sequences of the eruption in six passbands are provided in the supplementary movie . in order to clearly display the rising motion of the filament ,
a slice - time plot was constructed with base - difference images of 304 along the dotted line in panel b of figure 1 , as presented in figure 2 .
around 12:16:00 ut , the filament apex moved out of the field of view ( fov ) of aia , so no further kinematic evolution analysis was possible after that time .
through the time - stacking plot , we measure the filament height with time for a careful kinematic analysis as described in next subsection .
the kinematic information of the filament is obtained by analyzing aia 304 base - difference images .
we carefully inspect the images and identify the filament leading edges along the slice in figure 1(b ) .
the heights are measured from the projected distance of the leading edges from the flare location .
the uncertainty of the height measurement is about 4 pixels ( 2 mm ) , which are propagated to estimate the velocity errors in the standard way .
based on the height - time measurements , the velocities are derived from a numerical derivative method that employs lagrangian interpolation of three neighboring points , a piecewise approach to calculate the velocity ( zhang et al .
2001 ; cheng et al . 2013
; song et al
. 2014a , 2014b ) .
then the filament acceleration is calculated with the same method based on the calculated velocity profile .
figure 3 presents the kinematic analysis results .
the entire eruption process as seen by aia can be divided into three distinct phases as shown in panel a : an initial slow - rise phase , and then two fast acceleration phases .
the vertical green solid lines on the left and right demarcate the start time of the first and second fast acceleration phases , respectively .
in addition , the sxr flux with time can also be divided into a slow - rise phase and two impulsive phases .
the vertical black dotted line denotes the onset of the first impulsive phase of x - ray flare .
apparently , the right green solid line also marks the onset of the second impulsive phase of the flare .
the slow - rise phase of the filament lasted for about 7 min from @xmath212:00 to @xmath212:07 ut ; at the end of this phase , the velocity increased to @xmath240 km s@xmath3 with an averaged acceleration of 95 m s@xmath4 .
the first fast acceleration phase started around 12:07:19 ut and ended around 12:11:31 ut when the velocity reached to @xmath2450 km s@xmath3 with an averaged acceleration of @xmath21626 m s@xmath4 .
then the velocity decreased slightly during the following @xmath21.5 minutes .
the second fast acceleration phase started around 12:13:00 ut .
as the filament moved out of the aia fov , we can trace a part of the second acceleration phase .
the final velocity we deduce is @xmath2700 km s@xmath3 around 12:15:40 ut .
therefore , the averaged acceleration for this phase is @xmath21600 m s@xmath4 .
the horizontal red solid lines in figure 3 depict these two fast acceleration periods .
as mentioned , the sxr flux with time has two impulsive phases , which indicate there exist two obvious reconnection processes .
this point is further confirmed by the derivation of the sxr flux with two peaks as shown in panel b. the two red dots in panels b and c depict the positions of two peaks , which show that the first obvious reconnection is weak compared with the second one .
the onset of cme acceleration phase often coincides well with the onset of accompanying flares ( zhang et al .
2001 ) . for our event
, it is obvious that the velocity and sxr flux profiles are tightly consistent during their slow - rise phase and second fast acceleration phase .
however , the onset time of the first fast acceleration phase of filament ( 12:07:19 ut ) is obviously ahead of the first impulsive phase of flare ( 12:09:19 ut ) for two minutes , which means the filament was accelerated before the first obvious reconnection onset .
thanks to aia s high cadence , such short difference is otherwise difficult to observe and measure .
further , we plot the hxr flux versus time in panel c , along with the filament acceleration profile . the error bars of acceleration are large and not shown here .
note the hxr rates fell after 12:05 ut when the thin attenuators moved into the detector fov , and it fell suddenly again before 12:18 ut when _ rhessi _ went into night .
we made two gaussian fitting for the hxr flux as presented with two thin blue solid curves in the panel .
we believe the two fitted gaussian shapes correspond to the two obvious reconnection processes according to the derivation of sxr flux in panel b. the observed hxr peaks in the middle of the two fitted peaks should be due to the overlapping effect of hxr emission from these two reconnections , which is supported by the accumulation of the two fitted curves as shown with the thick blue solid curve .
panel c shows that the first fast acceleration profile did not have any correlations with the hxr flux , even the acceleration continued to decrease after the reconnection onset .
they tightly correlated with each other only during the second fast acceleration process .
* therefore , we suggest that the first fast acceleration process was induced by the instability ( cheng et al . 2013 ) , while the second one was accelerated mainly through the subsequent magnetic reconnection ( priest & forbes 2002 ; lin et al .
* the aia 335 observations present a kink morphology during the eruption ( see figure 1(h ) ) , indicating that the filament might be accelerated through the mfr kink instability ( cheng et al .
our study suggests that both mechanisms have a comparable contribution to the cme acceleration , which is consistent with the mhd simulation results ( chen et al .
our observational results have significant theoretical implications . *
the event provides us an excellent opportunity to compare the contributions of different acceleration mechanisms .
* during the eruption , the filament first exhibits a slow - rise phase , followed by two fast acceleration phases . for the slow - rise phase
, the acceleration might be induced by the initial weak quasi - separatrix - layer ( qsl ) reconnection ( cheng et al . 2013 ) , i.e. , reconnections at the interface between filament and its surrounding corona .
the brightening as shown with arrows in figures 1(a ) and ( d ) indicated that there existed reconnection during that period ( cheng et al .
another possibility is that the slow - rise phase was due to the initial acceleration stage caused by the mfr kink instability .
it is not possible to distinguish their contributions at this phase .
however , for the following two fast acceleration phases , we suggest that they are attributed to the instability and the reconnection , respectively .
it can be seen that the first obvious reconnection did nt result in any considerable filament acceleration .
this might be explained by two possible factors : first , the reconnection is weaker compared to the second one according to their hxr peak values as mentioned above .
* this weak reconnection might be insufficient to accelerate the mfr as it can not weaken the tension force of the overlying magnetic loops fast enough , and may even lead to the mfr deceleration as calculated by lin & forbes ( 2000 ) and lin ( 2002 ) * ; second , it is likely that this reconnection is still the qsl reconnection , as told from the hxr data of _ rhessi_. the locations of the hxr sources are shown as black contours at 70% , 80% , and 90% of the maximum in the 25 - 50 kev band ( figures 1(b),(c),(e),(f ) ) .
the sources are mainly at the two footpoints of the filament during the first obvious reconnection ( figures 1(b),(e ) ) , which is consistent with the qsl reconnection .
the qsl reconnection can heat the filament ( as depicted with arrows in figures 1(b ) and ( e ) ) ( cheng et al .
2013 ) and produce high energy electrons that escape along two legs of the filament and produce hxr emission at the footpoints , but it does not contribute to accelerate filament . on the other hand ,
the hxr sources locate mainly between the two footpoints during the second obvious reconnection ( figures 1(c),(f ) ) , which indicates the reconnection mainly took place in the current sheet connecting the filament to the flare loop .
this reconnection can accelerate filament / mfr ( carmichael 1964 ; sturrock 1966 ; hirayama 1974 ; kopp & pneuman 1976 ; lin & forbes 2000 ; lin 2002 ; chen 2011 ) and produce energetic electrons that hit the flare loop to produce the _ rhessi _ hxr source .
if there was not the first obvious magnetic reconnection , and the second obvious reconnection took place quickly after the kink instability onset , we will not be able to distinguish the contributions from the instability and reconnection , and we should observe only one fast acceleration phase like most events .
therefore , we point out that the time difference between the instability onset and the subsequent reconnection in the current sheet should be long enough to distinguish their contributions .
we conjecture that this time difference is very short in most events , which might be one important reason that similar events with two fast acceleration phases are rare .
* a filament eruption associated with an m5.9 class flare was well observed by the aia at the southeast limb of the sun on 2014 august 24 , which presented two fast acceleration phases during the eruption and provided us a perfect opportunity to compare the contributions of different acceleration mechanisms of cmes / mfrs during a particular event . *
based on the detailed analysis of the relations between velocity ( acceleration ) and sxr ( hxr ) profiles , we suggest that the instability and magnetic reconnection make a major contribution during the first and second fast acceleration phases , respectively .
the averaged acceleration for the first phase is @xmath21626 m s@xmath4 , similar to that of the second phase ( @xmath21600 m s@xmath4 ) .
therefore , both instability and reconnection play a comparable role to accelerate the filament in this event , which is consistent with the mhd simulation results ( chen et al
. 2007b ) .
we summarize the two fast acceleration phases as follow : the instability takes place first in a catastrophic manner , then the mfr is accelerated by the lorentz force ( chen et al .
2007b ; amari et al .
the magnetic energy is mainly transformed into the kinetic energy of mfr . in the meanwhile ,
current sheet is bound to form as the eruptive mfr drags magnetic field lines outwards , which provides proper site for fast magnetic reconnection .
the subsequent reconnection produces a further acceleration of the mfr , i.e. , the second fast acceleration phase .
we thank the referee for his / her valuable comments that improved the manuscript .
sdo is a mission of nasa s living with a star program .
this work is supported by the 973 program 2012cb825601 , nnsfc grants 41274177 , 41274175 , and 41331068 .
j.z . is supported by us nsf ags-1249270 and nsf ags-1156120 . | filament eruptions often lead to coronal mass ejections ( cmes ) , which can affect critical technological systems in space and on the ground when they interact with the geo - magnetosphere in high speeds .
therefore , it is an important issue to investigate the acceleration mechanisms of cmes in solar / space physics .
based on observations and simulations , the resistive magnetic reconnection and the ideal instability of magnetic flux rope have been proposed to accelerate cmes .
* however , it remains elusive whether both of them play a comparable role during a particular eruption .
* it has been extremely difficult to separate their contributions as they often work in a close time sequence during one fast acceleration phase .
here we report an intriguing filament eruption event , which shows two apparently separated fast acceleration phases and provides us an excellent opportunity to address the issue . through analyzing the correlations between velocity ( acceleration ) and soft ( hard ) x - ray profiles
, we suggest that the instability and magnetic reconnection make a major contribution during the first and second fast acceleration phases , respectively .
further , we find that both processes have a comparable contribution to accelerate the filament in this event . |
in the past decade , the pseudopotential method with the local - density - approximation for the exchange - correlation functional has had tremendous success in the study of ground - state properties of nonmagnetic systems.@xcite however , its applications to magnetic systems have been very rare because of the difficulties in dealing with localized @xmath1 electrons within this approach .
this difficulty is due to the rather large overlap of the electron densities of core and valence electrons , which makes it necessary that the exchange and correlation interaction of core and valence electrons is taken into account properly,@xcite i.e. , a linearization of this term , which is commonly assumed in pseudopotential studies , is not acceptable . indeed , recent studies have shown that partial - core corrected pseudopotentials provide an accurate description of the ground - state properties of @xmath0 transition metals.@xcite the pioneering pseudopotential calculations of greenside and schlter@xcite for ferromagnetic bcc fe were unfortunately not very successful possibly due to convergence problems caused by the fact that a gaussian basis set was employed : the calculated equilibrium lattice constant and magnetic moment were significantly lower than those resulting from all - electron calculations .
recently , zhu , wang , and louie@xcite performed a mixed - basis pseudopotential calculation for the ground - state properties of fe and found good agreement with the results of all - electron calculations ; and cho and kang@xcite found that the ground - state properties of ni were described well even when a `` soft '' @xmath1 pseudopotential is used together with a plane wave basis set .
these two recent calculations showed the importance of the core - valence exchange and correlation interaction and that a reliable description of the structural and magnetic properties of @xmath0 transition metals is indeed provided by the _ ab initio _ pseudopotential method . in the present work ,
we calculate the ground - state properties of fe , co , and ni using the linear - augmented - plane - wave ( lapw ) method and norm - conserving pseudopotentials .
the purpose of this study is to examine the accuracy of the lapw pseudopotential approach for magnetic 3@xmath1 transition metals in a systematic way .
since we employ the lapw basis which is efficient and accurate to represent localized as well as extended wave function , we can easily deal with the full core electron density ( used in the nonlinear core - valence exchange and correlation interaction ) as well as the @xmath1 valence electron density .
we find that the equilibrium lattice constant , bulk modulus , and magnetic moment calculated with both the local - spin - density approximation@xcite ( lsda ) and the generalized gradient approximation@xcite ( gga ) are in good agreement with those of all - electron calculations , respectively . compared to the lsda results ,
the gga results agree noticeably better with experiment . in agreement with previous studies@xcite
we find that the accurate treatment of the nonlinearity of the core - valence exchange and correlation interaction is essential for the proper description of the magnetic moment and magnetic energy .
we show that the lapw pseudopotential approach is an efficient and highly accurate method to calculate the ground - state properties of magnetic @xmath0 transition metals .
we will compare its advantages and disadvantages to those of the conventional plane wave method . the remainder of the paper is organized as follows . in sec .
ii , the calculational method is described . in sec .
iii , we present the calculated ground - state properties of fe , co , and ni and compare them with previous theoretical results and experiments . finally , a summary is given in sec .
the spin - polarized electronic - structure calculations presented in this work are performed using the lapw method@xcite and norm - conserving pseudopotentials@xcite within both the lsda and the gga .
we use the ceperley - alder@xcite ( ca ) and perdew - wang@xcite ( pw91 ) exchange - correlation functionals for the lsda and gga calculations , respectively .
the nonlocal ionic pseudopotentials of fe , co , and ni are generated from the ground - state atomic configuration by the generalized norm - conserving pseudopotential scheme of hamann.@xcite in the present lsda and gga calculations we use lsda pseudopotentials employing the ca exchange - correlation functional .
for comparison with the gga results from lsda pseudopotentials , we also perform the gga calculations for fe using gga pseudopotentials with the pw91 exchange - correlation functional .
the nonlinear core - valence exchange and correlation interaction is treated accurately by using the full core electron density which is computed in the atomic calculation . based on the lapw method ,
we expand the wave functions , electron density , and potential in terms of spherical harmonics inside the muffin - tin ( mt ) spheres and in terms of plane waves in the interstitial region .
the mt sphere radius is chosen to be @xmath2 = 2.1 bohr for all elements considered in this paper .
for the wave functions we employ spherical harmonics with an angular momentum up to @xmath3 = 10 and plane waves up tp a kinetic energy cutoff of @xmath4 = 13 ry .
for the electron density and potential we use @xmath5 = 6 and @xmath6 = 100 ry .
the electron density is obtained from the wave functions calculated at 44 and 47 @xmath7 points in the irreducible part of the brillouin zone for the bcc and fcc structures , respectively .
these calculational parameters are found to yield a well converged result for the ground - state properties of fe , co , and ni .
we start with a discussion of the nonmagnetic ( nm ) and ferromagnetic ( fm ) phases of bcc fe , fcc co , and fcc ni using the ca lsda functional.@xcite the calculated lattice constant , bulk modulus , and magnetic moment for the fm phase of fe , co , and ni are given in table i together with those of previous calculations@xcite and experiments.@xcite note that these other calculations employed different exchange - correlation functionals .
our results , when using the lsda ( ref .
10 ) , show a discrepancy for the equilibrium lattice constant , bulk modulus , and magnetic moment of about @xmath83 % , + 35 % , and @xmath810 % with respect to the experimental data , respectively .
these errors for 3@xmath1 transition metals are typical for lsda calculations .
as shown in table i , the present lsda results are in good agreement with all - electron lsda calculations such as the linear - combination - of - atomic - orbitals@xcite ( lcao ) and linear - muffin - tin - orbital@xcite ( lmto ) methods .
we note that the bulk modulus calculated by our lapw pseudopotential approach is systematically smaller than that of other calculations .
we believe that this is possibly due to the use of frozen - core pseudopotentials and different techniques for solving the one - electron kohn - sham equation . as in previous lsda
calculations,@xcite the present lsda study for fe fails to predict the experimentally observed ground state which is the fm bcc structure .
figure 1 shows our results fitted by murnaghan s equation of state.@xcite the nm fcc phase is energetically favored over the fm bcc phase .
the nm bcc phase is found higher in energy than the nm fcc phase by @xmath9 = 24.6 mry , and the fm bcc phase is higher than the nm fcc phase by @xmath10 = 4.4 mry . comparing these energy differences to other calculations ( see table ii )
we find good agreement : the full - potential linear - augmented - plane - wave@xcite ( flapw ) results were @xmath9 = 25.8 mry and @xmath10 = 4.1 mry , and the mixed - basis pseudopotential calculations@xcite obtained @xmath9 = 27 mry and @xmath10 = 5 mry .
as it is well known , some ground - state properties of @xmath0 transition metals are not correctly predicted by the lsda . in particular for fe , it has been demonstrated that nonlocal contributions to the exchange and correlation potential are necessary to predict the correct ground state.@xcite the previous work employed the pw86 gga functional.@xcite in the present work , we perform additional total - energy calculations for the ground - state properties of fe , co , and ni using the pw91 gga functional which fulfills all known sum rules best.@xcite for these gga calculations , we use lsda pseudopotentials calculated with the ca exchange - correlation functional .
the fitted energy - volume curves for various phases of fe are shown in fig.2 and the equilibrium lattice constant , bulk modulus , and magnetic moment of fe , co , and ni are given in the parentheses of table i. we find that the gga provides a substantial improvement over the lsda .
the increase in the lattice constant and magnetic moment , and the decrease in the bulk modulus lead to a better agreement with the experimental values . in the case of fe ,
our gga calculations predict correctly a fm bcc ground state ( see fig.2 ) .
we find that the nm bcc phase is higher in energy than the nm fcc phase by @xmath9 = 23.1 mry , but the fm bcc phase is lower than the nm fcc phase by @xmath10 = @xmath814.9 mry .
the magnetic energy @xmath11 ( i.e. , the energy difference between the nm and fm phases ) is thus 38.0 mry .
these energetics of fe compare well with previous calculations@xcite ( see table ii ) .
we see that our values are close to those of the all electron flapw@xcite calculation ( @xmath9 = 21.2 mry , @xmath10 = @xmath813.9 mry , and @xmath11 = 35.1 mry ) . on the other hand , the pseudopotential calculation of zhu , wang , and louie@xcite gave @xmath9 = 23 mry , and @xmath10 = @xmath823 mry .
the latter value is somewhat larger than both the all electron flapw and our present value .
zhu , wang , and louie pointed out that the overestimation of the magnetic energy in their gga calculations was due to the poor transferability of pseudopotentials within the gga functional .
in fact , a larger lattice constant with a reduction in the bulk modulus was found in their gga results ( see table i ) .
to check the transferability of gga pseudopotentials , we calculate the ground - state properties of fe using a gga pseudopotential with the pw91 exchange - correlation functional , i.e. , we use the pw91 functional in the atomic kohn - sham equation , in the unscreening of the effective potential of the atom to determine the ionic pseudopotential , and in the solid - state calculation .
we find that the ground - state properties of fe change little by using gga pseudopotentials .
the calculated lattice constant ( @xmath12 = 5.39 bohr ) , bulk modulus ( @xmath13 = 1.71 mbar ) , and magnetic moment ( @xmath14 = 2.30 @xmath15 ) are very close to those of our gga calculations with lsda pseudopotentials ( @xmath12 = 5.40 bohr , @xmath13 = 1.69 mbar , and @xmath14 = 2.32 @xmath15 ) ; and the energetics for the nm bcc , nm fcc , and fm bcc phases ( e.g. , @xmath9 = 23.3 mry , @xmath10 = @xmath814.2 mry , and @xmath11 = 37.5 mry ) are also in good agreement with the above lsda pseudopotential results .
we show that in the case of fe , both lsda and gga pseudopotentials yield very similar results when used together with a gga calculation for the valence electrons .
thus we speculate that the somewhat larger lattice constant and magnetic energy of the mixed - basis pseudopotential calculations@xcite may be attributed to the basis incompleteness and in particular to an inaccurate treatment of core - valence overlapping within the mixed basis set .
finally we investigate the importance of the nonlinearity of the exchange - correlation interaction of the core and valence electrons on the ground - state properties of fe , co , and ni .
the above calculations were done correctly , but we now linearize the exchange - correlation functional with respect to the interaction between the core and the valence electron density .
such a linearization is not necessary , but it is rather common in pseudopotential calculations . the calculated lattice constant , bulk modulus , magnetic moment , and magnetic energy are compared with the correct results in table iii .
the importance of a correct treatment of the exchange - correlation functional obvious .
in particular , there is a significant difference in the ground - state properties of fe . on the other hand
, we note that the lattice constant and bulk modulus of ni are affected only little , but the change in the magnetic energy @xmath11 is substantial ( see table iii ) .
hence we conclude that an accurate treatment of the nonlinear core - valence exchange and correlation interaction is essential for a trustworthy description of structural , elastic , and magnetic properties of 3@xmath1 transition metals .
we studied the ground - state properties of fe , co , and ni using the lapw method and norm - conserving pseudopotentials within both the lsda and the gga .
we found that the equilibrium lattice constant , bulk modulus , and magnetic moment are in good agreement with previous all - electron calculations .
compared to the lsda results , the gga results show a better ground - state properties of fe , co , and ni . we also find that the accurate treatment of the nonlinear core - valence exchange and correlation interaction is essential for the determination of the magnetic properties of @xmath0 transition metals .
the present results demonstrate well the reliability of the pseudopotential approach in describing the ground - state properties of magnetic @xmath0 transition metals .
there are several advantages for using the present combination of the lapw technique and pseudopotential approach .
this method provides an efficient and accurate treatment of 3@xmath1 transition metals , for which the localized @xmath1 electrons are difficult to handle in a conventional plane wave basis set . for a plane wave calculation a gga pseudopotential gets rather `` hard '' , but such hardness presents no difficulty in our lapw approach .
in particular , the use of pseudopotentials allows us to avoid inaccuracies of the lapw basis functions due to approximate orthogonalization between semicore and valence states , which is often adopted in the all - electron lapw method .
we thank p. alippi and g. vielsack for assistance in the initial stages of the calculations .
one of us ( j.h.c . ) would like to acknowledge the financial support from the korea science and engineering foundation .
m. l. cohen , science 234 , 549 ( 1986 ) .
w. e. pickett , comp .
phys . rep .
9 , 117 ( 1989 ) .
g. b. bachelet , d. r. hamann , and m. schlter , 26 , 4199 ( 1982 ) .
s. g. louie , s. froyen , and m. l. cohen , 26 , 1738 ( 1982 ) .
j. zhu , x. w. wang , and s. g. louie , 45 , 8887 ( 1992 ) .
cho and m .- h .
kang , 52 , 9159 ( 1995 ) .
h. s. greenside and m. schlter , 28 , 535 ( 1983 ) .
p. hohenberg and w. kohn , phys .
136 , b864 ( 1964 ) ; w. kohn and l. j. sham , phys .
140 , a1133 ( 1965 ) .
u. von barth and l. hedin , j. phys .
c5 , 1629 ( 1972 ) .
d. m. ceperley and b. j. alder , 45 , 566 ( 1980 ) ; j. p. perdew and a. zunger , 23 , 5048 ( 1981 ) .
d. c. langreth and m. j. mehl , 47 , 446 ( 1981 ) ; c. d. hu and d. c. langreth , phys .
32 , 391 ( 1985 ) .
j. p. perdew and w. yue , 33 , 8800 ( 1986 ) ; j. p. perdew , @xmath16 33 , 8822 ( 1986 ) ; 34 , 7406(e ) ( 1986 ) .
j. p. perdew in electronic structure of solids 91 , edited by p. ziesche and h. eschrig ( akademie - verlag , berlin , 1991 ) .
j. p. perdew , j. a. chevary , s. h. vosko , k. a. jackson , m. r. pederson , d. j. singh , and c. fiolhais , 46 , 6671 ( 1992 ) .
p. blaha , k. schwarz , p. sorantin , and s. b. trickey , comp .
59 , 399 ( 1990 ) ; wien93 , tu wien ( 1993 ) .
b. kohler , s. wilke , m. scheffler , r. kouba , c. ambrosch - draxl , to be published in comp .
d. r. hamann , 40 , 2980 ( 1989 ) .
t. c. leung , c. t. chan , and b. n. harmon , 44 , 2923 ( 1991 ) .
m. krling and j. hglund , 45 , 13293 ( 1992 ) .
d. j. singh , w. e. pickett , and h. krakauer , 43 , 11628 ( 1991 ) .
c. kittel , _ introduction to solid state physics _ , 6th edition ( john wiley , new york , 1986 ) .
o. gunnarsson and b. i. lundqvist , 13 , 4274 ( 1976 ) .
f. d. murnaghan , proc .
u.s.a . 3 , 244 ( 1944 ) .
. calculated lattice constant ( @xmath12 ) , bulk modulus ( @xmath13 ) , and magnetic moment ( @xmath14 ) for the ferromagnetic phase of fe , co , and ni in comparison with previous calculations and experiments .
the values are from the lsda calculation ( the gga results are in the parentheses ) .
abbreviations for the lsda and the gga represent the used exchange and correlation functionals [ ca = ceperley and alder ( ref . 10 ) , gl = gunnarsson and lundqvist ( ref .
22 ) , vbh = von barth and hedin ( ref .
9 ) , pw86=perdew and wang ( ref .
12 ) , pw91=perdew ( ref . 14 ) ] . [ cols="<,<,^,^,^",options="header " , ] | the ground - state properties of fe , co , and ni are studied with the linear - augmented - plane - wave ( lapw ) method and norm - conserving pseudopotentials .
the calculated lattice constant , bulk modulus , and magnetic moment with both the local - spin - density approximation ( lsda ) and the generalized gradient approximation ( gga ) are in good agreement with those of all - electron calculations , respectively .
the gga results show a substantial improvement over the lsda results , i.e. , better agreement with experiment .
the accurate treatment of the nonlinear core - valence exchange and correlation interaction is found to be essential for the determination of the magnetic properties of @xmath0 transition metals .
the present study demonstrates the successful application of the lapw pseudopotential approach to the calculation of ground - state properties of magnetic @xmath0 transition metals . |
during the compilation of the hipparcos input catalogue ( hic ) , special attention was given to obtain a uniform all sky coverage of late type variables , including the poorly known ones in the southern hemisphere .
the inclusion in hic of long period variables ( lpvs ) was possible only for those brighter than hipparcos magnitude hp 12.5 ( the detection threshold ) at least 80 % of their cycle .
one of prerequisites of the hipparcos mission was that the brightness of the targets needed to be known in advance to allocate the appropriate observing time .
however lpvs are not strictly periodic in their amplitudes , phases , and even periods .
thus the prediction of the brightness and of the observability windows ( time intervals when hp was @xmath012.4 ) could not be achieved without performing complementary ground - based observations before and during the mission on about 340 lpvs .
the responsibility of monitoring the hipparcos lpvs was taken by the aavso , both by continuing long term observations and adding new variables to the aavso observing programme ( mattei 1988 ) .
about one million longterm aavso observations , together with about 70000 yearly continuing observations were used to prepare and refine the ephemerides produced by the variable star coordinator at montpellier , france in collaboration with the aavso .
the main - mission photometry was performed in the wide hp band extending from 380 to 900 nm . due to irradiation by energetic solar and cosmic particles ,
the transmission of the detection chain suffered a severe wavelength - dependent deterioration during the mission .
the standard hp system was re - defined for an epoch near mid - mission . a photometric reduction to a subset of 22000 standard stars allowed to fix the zero point of the hp magnitude scale better than 0.001 mag twice per day .
the very red stars were a difficult case for the reduction to a standard hp system .
the reddest non - variable standards have v - i colour less than 1.8 whereas the mean v - i of most lpvs lies in a range from 2 to 6 magnitudes . note that the dominant flux for late type carbon stars is emitted in the 700 to 900 nm domain .
the early reduction algorithms were polynomial relations between the instrumental magnitudes and the standard hp@xmath1 as function of b - v .
this approach failed to model the aging effects for late type stars , especially m and s type giants , inducing spurious long term drifts and short term flickering when the reduction relations were extrapolated to red variables .
the technique to accurately define the chromatic aging from red variables , i.e. @xmath2hp versus v - i as a function of time , was accomplished by forcing hp - v@xmath3 to be constant , at given light - curve phase and v@xmath3 , throughout the mission .
the @xmath2hp /(v - i ) relation is a non- linear function of v - i . for reduction purposes ,
a linear pseudo - index was defined as given in table 1.3.2 of the hipparcos and tycho catalogues ( hip ) , vol .
1 ( esa 1997 ) .
this linearization procedure is nearly exact for stars with v - i less than 2 , but may leave reduction residuals of the order of few percent on individual magnitudes .
simultaneous observations , visual by aavso observers and photoelectric , with ccd camera and classical geneva photometer , were performed to tie the hp , the @xmath4 and @xmath5 , @xmath6 scales for red semiregular and mira type lpvs . for m ,
s , c stars a unique relation exists between hp and the johnson @xmath6 , as function of v - i .
note that m and c stars have a very distinct behaviour ( hp - v)/(b - v ) or ( hp-@xmath7)/(@xmath8-@xmath9 plane , see fig .
1.3.7 in the hipparcos and tycho catalogues ( hip ) , vol .
1 ( esa 1997 ) . the adopted relation hp - v versus v - i from the cousin s system ,
is given below : + [ cols="<,<,>,>,>,>,>,>,>,>,>",options="header " , ]
all r , n , c type stars observed by hipparcos turned out to be variable either irregular , semi - regular or nearly periodic .
since @xmath10 is larger than @xmath11 , the amplitude in the hp band is generally smaller than that in b and v bands .
this behaviour is illustrated in fig.1 for hip 59844 , the pulsating star bh cru .
[ vjhp_59844 ] the ratio q = @xmath12 is around 0.7 for early type c and m giants .
it shows a slight decrease for the reddest stars , see fig.2 .
[ q_ahv ] large amplitude periodic variables often show nearly sine folded light - curves , e.g. hip 109089 in fig.3 .
[ f_curves ] the bump near phase 0.7 is present in many light - curves as shown for hip 106583 , 99653 or 26753 .
semi - regulars and stars pulsating on the first overtone show rather noisy folded light curves due to the their varying amplitude and not so periodic behaviour . for small amplitude irregulars ,
the time coverage during the hipparcos mission was sufficient to describe their behaviour in terms of peak to peak amplitudes and time scales for variations , cf .
eyer et al.(1997 ) .
the hp amplitudes show two regimes for periodic r , n , c variables .
r stars and part of c , n stars show a linear relation between the period and the amplitude which may be expressed as @xmath13 = 1.3e-03@xmath14period . for the classical c - rich mirae with typical @xmath13 in the range 1.2 to 2.4 mag
, the amplitude shows little if any dependence on the period , ranging between 200 and 480 days .
the global behaviour is displayed in fig.4 .
[ rnc_amp ]
the visual estimates are obtained by interpolating the brightness of the variable star using a series of reference stars of known magnitude in its field . the difference between the response of the eye and of the @xmath6 band leads to an offset between the visual and the photoelectric @xmath6 , proportional to the colour difference between the comparison stars and the red variable .
the monitoring by aavso observers generally produces a dense time coverage but because observations of different observers are combined , and because the accuracy of the individual observation is only between 0.1 and 0.3 , the light curves are noisy .
thus , average light curves were obtained by fitting curves to the individual observations by fourier , polynomial or quintic splines methods .
these fitted curves were then transformed to hp magnitudes and hipparcos photometry was then superimposed to them .
the difference hp - v is a function of the star s temperature and of the circum- and inter- stellar extinction .
since the colour change as function of the phase is generally unknown , the technique to reduce aavso magnitudes to hp is to plot @xmath15 versus @xmath16 .
for lpvs the relation is often s - shaped as shown for the c0ev type star hip 4284 , see fig.5 .
[ vh_4284 ] the fine structure of @xmath15/@xmath16 diagrams depends mainly on the teff and log g variations and on the corresponding absorption changes due to tio , vo or cn , c2 , sic molecular bands .
emission lines and dust extinction introduce departures from the mean relation .
although distinct relations seem to exist for rising and falling parts of the light - curve , a unique third degree polynomial was used to transform fitted aavso magnitudes to hp magnitudes with an uncertainty of 0.1- 0.2 mag in most cases.this uncertainty is generally small compared to the peak - to - peak hp amplitude .
errors on comparison stars magnitudes are automatically corrected by this process .
the atlas of hipparcos - aavso light - curves , part b ( esa 1997 ) , contains 274 stars in common .
here we show few representative cases of c stars light - curves which would be difficult to interpret with hipparcos data alone .
this is especially true for rcb variable ry sgr , where the main minima were missed due to the peculiar hipparcos time sampling , and the semiregular variable v hya with a 530-day and over a 6000-day period where hipparcos observations were obtained while the star was slowly fading to the minimum of its longer period , see fig.6 .
[ l_curves ]
we sincerely thank variable star observers around the world whose dedicated observations provided vital support in the observations of lpvs .
we gratefully acknowledge the support of nasa under grant nagw-1493 which made it possible for the aavso to provide data support to the hipparcos mission and the swiss national science foundation for its support to activities at geneva observatory .
european space agency ( esa ) .
1997 , in _ the hipparcos and tycho catalogues _ , esa sp-1200 , vol . 1 and 12 , the netherlands .
+ eyer , l. and grenon m. 1997 , in _ hipparcos - venice 97 , 13 - 16 may , + venice , italy _ , esa sp-402 ( july 1997 ) , 467 .
+ mattei , j.a .
1988 , in _ scientific aspects of the input catalogue + preparation ii , sitges , 25 - 29 january _ , j. torra , c. turon eds . , 376 | accurate photometry was obtained for all programme stars during the 3.3-year hipparcos mission .
the final observing programme included several hundred mirae ( m ) , semiregular ( sr ) long - period and irregular ( l ) variables .
a detailed calibration of the aging of the optics allowed the evaluation of very precise magnitudes over the whole range of star colours .
since the time coverage of the satellite observations was not sufficient to describe the behaviour of m , sr , or l type variables , smooth curves were fitted statistically to the dense aavso observations .
these curves were then transformed to the hipparcos system in order to complement the hipparcos photometry and thus produce precise light curves with fuller time coverage , for a set of several hundred late - type variables , including most carbon stars brighter than v = 12.4 at minimum luminosity . a preliminary discussion of the behaviour of c stars , as observed from the space in the broad hp band , is given . |
this work was supported in part by the national science foundation grant no .
10874142 , 90921010 , and the national fundamental research program of china through grant no .
2010cb92304 , and the fundamental research funds for the central universities no . swjtu09cx078 . | motivated by the work [ phys .
rev . lett . 89 , 220404 ( 2002 ) ] for detecting the vacuum - induced berry phases with two - mode jaynes - cummings models ( jcms ) ,
we show here that , for a parameter - dependent single - mode jcm , certain atom - field states also acquire the photon - number - dependent berry phases after the parameter slowly changed and eventually returned to its initial value .
this geometric effect related to the field quantization still exists , even the filed is kept in its vacuum state . specifically , a feasible ramsey interference experiment with cavity quantum electrodynamics ( qed ) system is designed to detect the vacuum - induced berry phase .
pacs numbers : 42.50.ct , 03.65.vf gbksong in 1984 , berry showed that the state of a quantum system can acquire a purely geometric phase ( called now as berry phase ) , in addition to the usual dynamical phase , after slowly changed and eventually returned to its initial form @xcite .
basically , berry phase does not depend on the dynamical properties of the system , but just depends on the topological feature of the parameter space of the evolved system .
up to now , berry phase has been found in various systems , such as spins , polarized lights , atoms and so on @xcite .
also , recent studies have shown that the geometric phases can be utilized to implement quantum logic gates for realizing quantum computation @xcite .
quantized optical fields , as well as their interactions with atoms , are the main objects in quantum optics .
originally , the famous jaynes - cummings model ( jcm ) @xcite is introduced to describe the interaction between an undamped two - level atom and a non - decaying single - mode quantized field , under the rotating - wave approximation .
this model has been widely generalized to treat various interactions between atoms and photons .
these include , e.g. , multilevel atoms interact with multimode quantized fields , and various multiphoton processes in quantum optics @xcite .
one of the basic phenomena in quantum optics is that , the vacuum of a quantized field can behave as a physical reality with certain observable effects .
for example , in terms of vacuum fluctuations of the quantized electromagnetic field @xcite certain important quantum effects , such as lamb shifts and spontaneous emissions can be well explained .
recent works @xcite indicated that , the vacuum of quantized optical field could also induce the observable berry phases . in order to observe these vacuum - induced geometric effects , two filed modes were introduced @xcite to interact with a two - level atom . as a consequence , the experimental tests are relatively complicated . in this work ,
we show that only one field mode interacting with a two - level atom could be utilized to detect the vacuum - induced berry phase . beginning with a generic model , i.e. , the @xmath0-quantum jcm
, we show how the desirable berry phase can be acquired by an evolved quantum state in a parameter - dependent single - mode jcm .
furthermore , we design a ramsey interference device involved with only one filed mode to detect such a geometric effect related to the field quantization .
the hamiltonian of a @xmath0-quantum jcm @xcite , i.e. , a two - level system coupled to a quantized mode via a @xmath0-photon process , can be expressed as ( under the rotating - wave approximation ) @xmath1 here , @xmath2 and @xmath3 are the creation and annihilation operators of the cavity field with frequency @xmath4 .
@xmath5 , @xmath6 , and @xmath7 are the pauli operators of the atom .
@xmath8 is the transition frequency of the atom between the excited state @xmath9 and ground state @xmath10 , and @xmath11 the coupling coefficient between the atom and cavity mode . under a time - dependent unitary transformation @xmath12 ,
the above hamiltonian can be rewritten as @xmath13 where @xmath14 is the detuning .
the eigenstates of such a hamiltonian read @xmath15 and @xmath16 respectively . above , @xmath17 are the number states the quantized bonsic field , and @xmath18 $ ] . following fuentes - guridi _ et .
al . _
@xcite , we introduce a phase - shift operation @xmath19=\exp[-i\phi(t)a^{\dag}a]$ ] to change the hamiltonian ( 2 ) to the following parameter - dependent form @xmath20 obviously , such a @xmath21-dependent hamiltonian describes a two - level atom interacting ( via a @xmath0-photon process ) with a quantized field mode . here , the phase parameter @xmath22 changes with the time @xmath23 and can be changed slowly from @xmath24 to @xmath25 generating a cyclic path in the parameter space during the evolution . as a consequence ,
if the system begins with one of its eigenstates , @xmath26 or @xmath27 , then it returns to such state but acquires a geometric phase ( besides the dynamical one not shown here ) @xmath28 or @xmath29 it is seen that the geometric phase acquired here depends on the photon number @xmath30 .
physically , this nontrivial quantum effect can be measured by using an interference procedure between the eigenstate @xmath31 ( or @xmath31 ) and the ground state @xmath32 , for which no geometric phase is acquired .
typically , if the system begins with the state @xmath33 , i.e. , the field is in a vacuum state , the above adiabatic operation performed on the degrees of freedom of the field still yields a berry phase @xmath34 , respectively.,width=453,height=249 ] with the solid angle @xmath35 $ ] . fig .
1 shows how the above vacuum - induced berry phase varies with the parameter @xmath36 . through adiabatic evolution ,
the initial state @xmath37 coupling to vacuum mode in cavity acquires a geometric phase . in this case , the atom - field entanglement in the eigenstates ( 3 ) , ( 4 ) can not be neglected @xcite .
note that the expression ( 8) can not only be interpreted as a geometric phase of the two level system , as the origin of the geometric phase is related to the vacuum fluctuation of the field .
clearly , for a common @xmath38 the more quantum @xmath0 corresponds to the greater vacuum - induced berry phase . basically , the photon - dependent berry phase shown in eqs .
( 6 - 7 ) is due to the performance of the field quantization . thus , even the photon number of the field is @xmath24 , the geometric phase is still nontrivially induced .
any classical correspondence of such a phenomenon does not exist .
berry phases related to field quantization could be measured with the usual one - photon jcms , which had been experimentally demonstrated in the well - known cavity qed systems .
indeed , various quantum natures @xcite of the radiation field interacting with atoms have been successfully demonstrated with these systems .
typically , a cavity qed experiment @xcite involved with _ two _ quantized bosonic modes was proposed to test the geometric phases generated in a two - mode jcm .
below , we show that a cavity qed system involved only one quantized bosonic mode could also be utilized to test the above vacuum - induced berry phase .
our proposed setup for such a test is shown in fig . 2 , wherein an atom is emitted from the source @xmath39 and then flies sequentially across @xmath40 , @xmath41 and @xmath42 , and is finally detected in @xmath43 . , and flies sequentially across the first ramsey zone @xmath40 , high - q quantized cavity @xmath41 ( wherein the cavity is kept in the vacuum state and the desirable vacuum - induced berry phase is generated by a classical driving from @xmath44 ) , the second ramsey zone @xmath42 , and then is finally detected in @xmath43 .
the information of the vacuum - induced berry phase is extracted by the measured atomic probability.,width=585,height=226 ] initially , the atom is assumed to be prepared in the upper level @xmath37 in the source @xmath39 and then emitted .
after the first ramsey zone @xmath40 , the state of the atom reads @xmath45 being the time spent by the atom inside the zone @xmath46 . during the atom flies across the high - q quantized cavity @xmath41 , the parameter - dependent hamiltonian ( 5 ) can be obtained .
for example , a raman configuration shown in fig .
3 is utilized to achieve the @xmath21-dependent one - photon jcm . here , an auxiliary external classical laser beam @xmath47 is applied to drive the transition @xmath48 with the rabi frequency @xmath49 , while the quantized cavity mode ( @xmath50 ) couples to the transition @xmath51 with the strength @xmath52 .
the hamiltonian describing such a configuration in the interaction picture reads ( @xmath53 ) ( see , e.g. , @xcite ) @xmath54 with @xmath55 being the detuning .
generally , the corresponding time - evolution operator can be formally expressed as @xmath56 under the so - called large - detuning limit , i.e. , @xmath57 , the second - order contribution to @xmath58 is significantly important than the first order one .
this is because that the former involves terms linear in time , whereas the latter involves the terms that are just oscillatory or constant in time .
therefore , we can only retain the above second - order terms and rewrite the above time - evolution operator ( 14 ) as @xmath59\right\}t=1-ih_{\rm eff}t,\ ] ] with an effective hamiltonian @xmath60,\,\lambda_1=\frac{\omega _ { 0}g}{\delta}.\ ] ] obviously , this effective hamiltonian is equivalent ( apart from the unimportant stark shifts ) to the above @xmath21-dependent hamiltonian ( 5 ) with @xmath61 .
therefore , after passing the cavity vacuum wherein the driving parameter @xmath21 changes from @xmath24 to @xmath25 , the atom undergoes the following evolution @xmath62 with rabi frequency @xmath52 .
in addition , a classical laser field driving the transition @xmath63 ( with rabi frequency @xmath64 ) is applied to produce the desirable @xmath21-dependent single - mode jcm.,width=340,height=249 ] , the probability of the atom being detected in the state @xmath37 . this probability is a function of the controllable parameter @xmath65 related to the atom interacting with the cavity . here
, the blue curve corresponds to the ramsey interferometry without berry phase , while the red curve shows the situation in which a geometric phase shift ( @xmath66 ) is induced .
additionally , the black ( dashed ) curve shows that in the presence of the cavity decay @xmath67 khz , the @xmath68 is little influenced.,width=510,height=325 ] here , @xmath69 is the vacuum - induced berry phase acquired in the pass of the cavity , and @xmath70 with @xmath71 being the duration of the atom stayed in the cavity .
furthermore , after passed the second ramsey zone @xmath42 , the atom evolves to the state @xmath72 with @xmath73 if the two ramsey zones are properly set such that the condition @xmath74 is exactly satisfied , then the probability of detecting the atom in its upper level @xmath37 in @xmath43 is @xmath75 if @xmath76 is set inside the cavity , the above probability can be further simplified to @xmath77 which is directly related to the berry phase acquired by the atom flying across the high - q quantized cavity @xmath41 .
therefore , berry phase generated in the one - photon jcm could be observed by the above ramsey interference method . experimentally , the above one - photon rabi frequency is set as @xmath78khz @xcite .
this implies that , if the solid angle is required as @xmath79 , then the parameter @xmath80 should be set to satisfy the condition @xmath81 .
since the parameter @xmath80 is determined above by @xmath82 $ ] , with @xmath83 , the rabi frequency of the applied classical driving should be designed as @xmath84 . on the other hands , in order to satisfy the large detuning condition required above , i.e. , @xmath85 , we may typically set @xmath86 yielding @xmath87 .
this means that the atom - field interaction cam perform @xmath88 complete rabi cycle during an effective atom - cavity interaction time of @xmath89 ms @xcite .
this interaction time is manifestly shorter than the decaying time ( @xmath90ms ) of the cavity ( see , e.g. , @xcite ) .
we now discuss how the dissipation of the cavity influence on the observable effect of the vacuum - induced geometric phase . following the ref .
@xcite , the effective hamiltonian of the atom - cavity system becomes : @xmath91 . with the same procedure we can prove that , through a cyclic and adiabatic evolution the acquired geometric phase reads @xmath92
since @xmath93 should be a perturbation quantity , we can expand the above geometric to the second order in @xmath94 and then obtain @xmath95 consequently , the probability of the atom being detected in the state @xmath37 is changed to be @xmath96/2 $ ] .
notice that in the case of low decoherence , the lowest order correction of the expected geometric phase is only quadratic in @xmath93 , suggesting that the field decoherence may not play such an important role in the proposed experiment .
this can be numerically verified from the comparison in the fig . 4 : after considering the presence of a typical cavity dissipation @xmath97 @xcite , the probability of the atom being detected in the state @xmath37 ( the black line ) is almost unchanged .
this means that the experimental detection of the vacuum - induced berry phase in jcm with the above ramsey interference is feasible , even in presence of the cavity losses . in summary
, we have calculated the berry phase of @xmath0-quantum jcm and proposed an experimental setup to observe and measure such a geometric phase induced by the vacuum field in an one - photon single - mode jcm . basically , geometric phases acquired by the atom - field system are dependent of the number of photons in the field .
this is different from those attained in semi - classical counterpart .
our results show also that , for a common @xmath38 , the more quantum @xmath0 corresponds to the larger vacuum - induced berry phase .
a ramsey interference experiment with cavity qed is designed to detect the vacuum - induced berry phase . |
in trying to elucidate the role of quarks in hadrons and nuclei one steps immediately into the domain of relativistic quantum field theory where within the key phenomena can only be understood via nonperturbative methods .
two prime examples are : confinement , the empirical fact that quarks have not hitherto been detected in isolation ; and dynamical chiral symmetry breaking , which is responsible , amongst many other things , for the large mass splitting between parity partners in the spectrum of light - quark hadrons , even though the relevant current - quark masses are small .
neither of these phenomena is apparent in qcd s lagrangian and yet they play a dominant role in determining the observable characteristics of real - world qcd .
the physics of hadrons is ruled by such
_ emergent phenomena_.
in connection with confinement it is worth emphasising at the outset that the potential between infinitely - heavy quarks measured in numerical simulations of quenched lattice - regularised qcd the so - called static potential is simply not relevant to the question of light - quark confinement .
in fact , it is quite likely a basic feature of qcd that a quantum mechanical potential between light - quarks is impossible to speak of because particle creation and annihilation effects are essentially nonperturbative .
a perspective on confinement was laid out in ref.@xcite .
expressed simply , confinement can be related to the analytic properties of qcd s schwinger functions , which are often loosely called euclidean - space green functions . for example
, it can be read from the reconstruction theorem that the only schwinger functions which can be associated with expectation values in the hilbert space of observables ; namely , the set of measurable expectation values , are those that satisfy the axiom of reflection positivity @xcite .
this is an extremely tight constraint .
it can be shown to require as a necessary condition that the fourier transform of the momentum - space schwinger function is a positive - definite function of its arguments .
however , that is not sufficient . in relation
to @xmath3-point schwinger functions , which are those connected with the propagators of elementary excitations in qcd , the axiom of reflection positivity is satisfied if , and only if , the schwinger function possesses a klln - lehmann representation .
this statement is most easily illustrated for a scalar field , in which case it means that one can write the scalar - field s @xmath3-point function in the form , @xmath4 ; and @xmath5 . a timelike vector , @xmath6 , has @xmath7 .
naturally , no theory consistent with causality can produce a schwinger function with a pole at spacelike @xmath8 . ]
@xmath9 the spectral density for a non - interacting scalar field of mass @xmath10 is @xmath11 .
it is plain that no function @xmath12 which falls - off faster than @xmath13 at large spacelike momenta can be expressed in this way .
an appreciation of the importance of the axiom associated with reflection positivity has led to the formulation of a _ confinement test _ @xcite . with a momentum - space schwinger function , @xmath14 , in hand , the first step is to calculate @xmath15 which gives the configuration - space schwinger function in the rest frame .
one then examines the properties of @xmath16 .
if an asymptotic state of mass @xmath10 is associated with this schwinger function , then @xmath17 i.e. , the schwinger function is positive definite and the mass of the asymptotic , propagating state is given by @xmath18 if , on the other hand , @xmath16 calculated from a particular schwinger function is not positive definite , then the axiom of reflection positivity is violated and the associated elementary excitation does not appear in the hilbert space of observables .
thus the appearance of a least one zero in @xmath19 is a sufficient condition for confinement .
it is a very clear signal .
an exemplar is provided by @xmath20 a function for which with any particular choice of @xmath21 there exists a @xmath22 such that @xmath23 one has @xmath24 .
one calculates from eq.([stingl ] ) : @xmath25.\ ] ] this model was proposed in ref.@xcite and argued therein to describe a confined gluon because it does nt possess a pole on the real-@xmath8 axis .
the argument for confinement is supported by the fact that @xmath16 is not positive definite and , indeed , exhibits damped oscillations about zero . hence the excitation associated with this schwinger function can not be connected with an expectation value in the hilbert space of observables .
the existence of a zero in @xmath16 and its disappearance have been advocated as a means by which deconfinement can be studied in qcd at nonzero temperature and density @xcite .
deconfinement and chiral symmetry restoration are coincident in all self - consistent studies of concrete models of qcd that exhibit both phenomena in vacuum .
this result appears to follow from the crucial role played by the in - medium evolution of the dressed - quark self - energy in both transitions .
further studies are needed to fully elucidate the nature and the probable simultaneity of these transitions in qcd .
it was observed and emphasised in ref.@xcite that a violation of reflection positivity is a sufficient but _ not _ necessary condition for confinement .
this criterion is nevertheless a powerful discriminating tool , as can be illustrated through a concrete example , which also serves to introduce the dyson - schwinger equations ( dses ) .
the dses provide a nonperturbative approach in the continuum for calculating and modelling the properties of schwinger functions .
they have long been used to develop insight into gauge field theories at both weak and strong coupling @xcite . in connection with confinement and dynamical chiral symmetry breaking ( dcsb )
the dse for the dressed - quark propagator is of fundamental importance . in qcd
this equation takes the form @xmath26 where at this point it is sufficient to work with the unrenormalised equation . in eq.([gendseur ] ) , @xmath27 and @xmath28 are the dressed - gluon 2-point function ( propagator ) and dressed - quark - gluon 3-point function ( vertex ) ; and @xmath10 is a current - quark bare mass .
finally , the subject of the equation , @xmath29 , is the dressed - quark propagator in vacuum .
it can be written in a number of equivalent forms : @xmath30 each of which involves just two scalar functions . in the last representation they are chosen to be the wave function renormalisation , @xmath31 , and the running mass function , @xmath32 .
reference @xcite introduced a simple _ ansatz _ for the kernel of eq.([gendseur ] ) ; namely , @xmath33,\ ; \gamma^a_\nu(q , p ) = \frac{\lambda^a}{2}\gamma_\nu\,,\ ] ] where @xmath34 is a mass@xmath35-scale .
the ladder quark - antiquark interaction represented by eqs.([mnprop ] ) is a constant in configuration space .
therefore the usual proof of the cluster decomposition property fails and hence this _ ansatz _ defines a confining model .
furthermore , with @xmath36 , eqs.([gendseur ] ) and ( [ mnprop ] ) yield the following solution : @xmath37 when eq.([deltatau ] ) is evaluated with @xmath38 constructed from eqs.([abmnres ] ) , one obtains @xmath39 i.e. , a function with repeated zeros .
thus the interaction s violation of the cluster decomposition property is expressed in a fermion propagator that fails to satisfy the axiom of reflection positivity .
hence it describes a confined elementary excitation .
the story continues in a fascinating manner when one dresses the quark - gluon vertex .
studies attempting to elucidate the nature of @xmath40 are now underway using both continuum @xcite and lattice - qcd techniques @xcite . however , this is a relatively recent development .
it was and remains common to employ an _
ansatz _ for this 3-point schwinger function that is constructed according to lessons learnt by studying ward - takahashi identities in gauge field theories @xcite .
one often used form is based on the vertex introduced in ref.@xcite ; viz .
, do not change the results which follow .
the influence of more significant changes ; for example , introducing a dependence on functions other than those which appear in the quark propagator , must be explored on a case - by - case basis .
it is a simple truth that all we have at present are _ anstze _ for the vertex and no modification will alter a simple , long established fact , soon to become apparent : the vertex can have a material impact on the expression of confinement and dcsb in qcd . ]
@xmath41 , with @xmath42 \!,\\
\sigma_f(\ell_1 ^ 2,\ell_2 ^ 2 ) & = & { \mbox{\footnotesize $ \displaystyle \frac{1}{2}$}}\,[f(\ell_1 ^ 2)+f(\ell_2 ^ 2)]\,,\ ; \delta_f(\ell_1 ^
2,\ell_2 ^ 2 ) = \frac{f(\ell_1 ^ 2)-f(\ell_2 ^ 2)}{\ell_1 ^ 2-\ell_2 ^ 2}\ , , \label{deltaf}\end{aligned}\ ] ] where @xmath43 ; viz . , the scalar functions in eq.([spab ] ) .
if one solves the gap equation , eq.([gendseur ] ) , using this vertex model and @xmath44 from eq.([mnprop ] ) one finds in the chiral limit @xcite @xmath45)]}{p^4}\,,\ ; \sigma_s(p^2 ) = \frac{1}{2\surd { \cal g}}\,\exp(-p^2/[2 { \cal g}])\,.\ ] ] it is apparent that dressing the vertex has markedly altered the form of the dressed - quark propagator ; namely , @xmath46 , which each had a branch point with the bare vertex , are transformed by the dressed vertex into entire functions .
moreover , when eq.([deltatau ] ) is evaluated with @xmath38 constructed from eqs.([vsmnres ] ) , one obtains @xmath47 in contrast to eq.([dabmn ] ) , this is positive definite .
however , it yields an effective mass via eq.([massdelta ] ) : @xmath48 and infinitely massive states _ can not _ propagate .
these observations notwithstanding , confinement of the quark represented by the propagator in eq.([vsmnres ] ) is still signalled by a violation of the reflection positivity axiom .
it is clear from the momentum space form of the propagator that the axiom is breached because both @xmath46 fall faster than @xmath13 at large spacelike @xmath8 .
however , it can also be seen in another way , which is indispensable if the momentum - space form of the schwinger function is not known .
one can evaluate eq.([deltatau ] ) with eq.([klrep ] ) to obtain @xmath49 reflection positivity is satisfied by a configuration - space schwinger function if , and only if , it can be expressed in this form , with @xmath50 as described in eq.([klrep ] ) .
now the theorem of ref.@xcite can be employed to show that the function in eq.([dgt2 ] ) can not be represented in this way and therefore contravenes the axiom .
it is a fact that no function which falls faster than @xmath51 at large @xmath52 , where @xmath53 is some constant , can be expressed through eq.([widder ] ) . amongst other things
, it follows that an excitation described by an effective mass , @xmath54 , which grows with euclidean separation is necessarily absent from the spectrum of physical states because its schwinger function violates the axiom of reflection positivity .
this is a novel perspective . with the information provided above one
can identify a confined excitation ; and hence , in the context of qcd , these tools have been employed to analyse propagators obtained both as solutions of truncated systems of dses and through numerical simulations of lattice - regularised qcd .
extant analyses , e.g. , ref.@xcite , suggest strongly that the gluon @xmath3-point function in qcd is not reflection positive . the same has been argued for the dressed - quark propagator in qcd , e.g. , ref.@xcite . however , as noted above , the detailed structure of the dressed - quark - gluon vertex is unknown and it can influence the analytic properties of @xmath29 .
hence a firm conclusion on the dressed - quark propagator has not yet been reached ( see , e.g. , ref.@xcite ) . in relation to the studies just mentioned
, it is noteworthy that any @xmath3-point schwinger function with an inflexion point at @xmath55 must breach the axiom of reflection positivity .
this can be seen directly from eq.([klrep ] ) , which constrains the first derivative of reflection positive functions to be negative definite and the second derivative to be positive definite .
hence , it is often unnecessary to actually calculate the fourier transform described in eq.([deltatau ] )
. a violation of reflection positivity can be determined by inspection of the pointwise behaviour of the momentum space schwinger function .
it should now be clear that we have a means of identifying whether or not a given schwinger function can be associated with an expectation value in the hilbert space of observables . in principle
, one could calculate all qcd s schwinger functions and demonstrate confinement of colour therewith .
plainly , however , that will be difficult in practice and nigh on impossible by brute force .
a proof of confinement at some higher organisational level is necessary , and sought by many .
there is a surviving hope that the study of gauge theories in less than four dimensions might be useful in this regard .
analogous to quenched qcd , quenched qed in three dimensions ( two spacial , one temporal
qed@xmath56 ) is confining because it has a nonzero string tension @xcite .
the effect of unquenching ; viz . , allowing light fermions to influence the theory s dynamics , remains an important but unresolved question .
it is conceivable that through its answer we might inform ourselves about aspects of the confinement problem in qcd .
the early history of qed@xmath56 studies is reviewed in ref.@xcite and their current status can be traced from ref.@xcite . in closing this section
it is important to reiterate that confinement is a conjecture .
in order to establish confinement as more than a contemporary empirical fact , it will be necessary to map out the nonperturbative structure of qcd s @xmath57-function .- function . for a detailed discussion , see , e.g. , ref.@xcite . ] in order to accomplish that , one will need to determine the nonperturbative evolution of the renormalisation constants .
that is tantamount to establishing rigorously that the theory exists .
this fact emphasises the magnitude of the problem .
absent this , progress in hadron physics can still be made with well - constrained models that express key consequences of confinement .
the feedback that this enables between experiment and theory further illuminates the essence of the confinement conjecture in relation to light - quarks .
this can begin with the development and employment of a poincar covariant framework that ensures the colour - singlet @xmath58-matrix elements associated with physical processes do not possess production thresholds for light - quarks , in particular , and coloured elementary excitations in general .
the notions described above have been immensely valuable in this regard . for example , they underly the dse model introduced in ref.@xcite , which has served as an archetype for numerous efficacious applications , as can be seen , e.g. , from refs.@xcite .
many statements of fact can be made in connection with this emergent phenomenon .
for example , dcsb explains the origin of constituent - quark masses and underlies the success of chiral effective field theory .
understanding dcsb within qcd proceeds from the renormalised gap equation @xcite : @xmath59 where @xmath60 represents a poincar invariant regularisation of the integral , with @xmath61 the regularisation mass - scale , @xmath27 is the renormalised dressed - gluon propagator , @xmath62 is the renormalised dressed - quark - gluon vertex , and @xmath63 is the quark s @xmath61-dependent bare current - mass .
the vertex and quark wave - function renormalisation constants , @xmath64 , depend on the gauge parameter .
the form of the solution to eq.([gendse ] ) is the same as that written in eq.([spab ] ) _ except _ that in qcd one must account for the renormalisation point dependence : @xmath65 it is important that the mass function , @xmath66 , illustrated in fig.[gluoncloud ] , is independent of the renormalisation point , @xmath67 . the dressed propagator is obtained from eq.([gendse ] ) augmented by the renormalisation condition . ]
@xmath68 where @xmath69 is the running mass : @xmath70 with @xmath71 the lagrangian - mass renormalisation constant . in qcd the chiral limit
is strictly defined by @xcite : @xmath72 which states that the renormalisation - point - invariant current - quark mass @xmath73 . in perturbation
theory it is impossible in the chiral limit to obtain @xmath74 : the generation of mass _ from nothing _ is an essentially nonperturbative phenomenon . on the other hand ,
it is a longstanding prediction of nonperturbative dse studies that dcsb will occur so long as the integrated infrared strength possessed by the gap equation s kernel exceeds some critical value @xcite .
there are strong indications that this condition is satisfied in qcd @xcite ( see fig.[gluoncloud ] ) .
it follows that the quark - parton of qcd acquires a momentum - dependent mass function , which at infrared momenta is @xmath75-times larger than the current - quark mass .
this effect owes primarily to a dense cloud of gluons that clothes a low - momentum quark @xcite .
it means that the higgs mechanism is largely irrelevant to the bulk of normal matter in the universe .
instead the single most important mass generating mechanism for light - quark hadrons is the strong interaction effect of dcsb ; e.g. , one can identify it as being responsible for 98% of a proton s mass .
the vacuum quark condensate is the order parameter most commonly cited in connection with dcsb .
for example , it occurs as a parameter at leading order in chiral perturbation theory .
it is a basic fact that only in the chiral limit is it possible to unambiguously define the gauge invariant vacuum quark condensate in terms of @xmath29 @xcite . that such a definition is possible at all emphasises that gauge covariant quantities contain gauge invariant information
notwithstanding these truths , @xmath32 is a more fundamental tracer for dcsb .
the condensate is only a small part of the information that the mass function contains , amongst other pieces are pointwise consequences of goldstone s theorem @xcite , definitions and an explanation of the constituent - quark mass @xcite , and a related @xmath76-term @xcite .- term and the massive - quark condensate defined unambiguously in ref.@xcite demonstrate that the essentially dynamical component of chiral symmetry breaking decreases with increasing current - quark mass .
this is an important observation that is basic , e.g. , to the validity of heavy - quark effective theory . ]
harking back to sect.[sect : conf ] , it is interesting to note that considered as a function of current - quark mass the dressed - quark propagator possesses a spectral representation @xcite , confinement and dcsb notwithstanding . that a nonzero quark mass function is generated in massless qcd
entails that chiral symmetry is realised in the nambu - goldstone mode and the vacuum quark condensate is nonzero .
the magnitude of the mass function in the infrared and order parameters contingent thereupon underly the validity of chiral effective theory as a tool for correlating low - energy observables in qcd . the question of whether @xmath32 has an expansion in current - quark mass around its chiral - limit value bears upon the radius of convergence for that perturbation theory .
this question asks whether it is possible to write @xmath77 on a measurable domain of current - quark mass .
it was found @xcite that such an expansion exists for @xmath78 and is absolutely convergent .
the value of @xmath79 can be reported as follows : is never within the domain of absolute convergence . ] for a pion - like meson constituted from a quark , @xmath80 , with mass @xmath79 and an equal - mass but different flavor antiquark , @xmath81 , @xmath82gev . since physical observables , such as the leptonic decay constant ,
are expressed via @xmath32 , it follows that a chiral expansion is meaningful only for @xmath83gev@xmath35 .
hence , it is only valid to employ chiral perturbation theory to fit and extrapolate results from numerical simulations of lattice - regularized qcd when the simulation parameters provide for @xmath84gev@xmath35 .
lattice results at larger pion masses are not within the domain of convergence of chiral perturbation theory .
dynamical chiral symmetry breaking is a fact in qcd and this amplifies the importance of the axial - vector ward - takahashi identity .
the existence of a sensible truncation of the dses @xcite has enabled proof via that identity of a body of exact results for pseudoscalar mesons .
they relate even to radial excitations and/or hybrids @xcite , and heavy - light @xcite and heavy - heavy mesons @xcite .
the results have been illustrated using a renormalisation - group - improved rainbow - ladder truncation @xcite , which also provided a prediction of the electromagnetic pion form factor @xcite .
in addition , algebraic parametrisations of the dressed - quark propagators and meson bound - state amplitudes obtained from such studies continue to be useful .
one example is a recent application to @xmath85-meson @xmath86 light - meson transition form - factors @xcite , the results of which can be useful in the analysis and correlation of the large body of data being accumulated at extant heavy - quark facilities , and thereby in probing the standard model .
implications for neutral pseudoscalar mesons have been elucidated @xcite . in the general case
the axial - vector ward - takahashi identity is written @xmath87 wherein : @xmath88 are the generators of @xmath89 ; the dressed - quark propagator @xmath90diag@xmath91 $ ] is matrix - valued ; @xmath92 is the matrix of renormalised ( running ) current - quark masses and @xmath93,$ ] where the trace is over flavour indices .
the inhomogeneous axial - vector vertex in eq.([avwti ] ) satisfies a bethe - salpeter equation ; viz . , @xmath94_{tu } = z_2 \left[\gamma_5\gamma_\mu { \cal f}^a \right]_{tu}+ \int^\lambda_q [ { \cal s}(q_+ ) \gamma^a_{5\mu}(q;p ) { \cal s}(q_-)]_{sr } k_{tu}^{rs}(q , k;p)\ , , \label{avbse}\ ] ]
where : @xmath95 is the total momentum and @xmath96 , @xmath97 are relative momenta , @xmath98 ; @xmath99, ,@xmath100 represent colour , dirac and flavour indices ; and the quantity @xmath101 is the bethe - salpeter kernel .
the pseudoscalar vertex @xmath102 satisfies an analogous equation driven by the inhomogeneity @xmath103 . the final term in eq.([avwti ] )
expresses the non - abelian axial anomaly .
it involves @xmath104 wherein the matrix element represents an operator expectation value in full qcd and @xmath105 is the topological charge density operator , with @xmath106 the gluon field strength tensor .
it is fundamentally important that while @xmath107 is gauge invariant , the associated chern - simons current , @xmath108 , is not .
thus in qcd no physical state can couple to @xmath108 .
hence , physical states can not provide a resolution of the so - called @xmath0-problem ; i.e. , they can not play any role in ensuring that the @xmath109 is not a goldstone mode .
it is plain and important that only @xmath110 can be nonzero .
moreover , in considering the @xmath0-problem one need only focus on the case @xmath111 because if that is false , then following ref.@xcite it is clear that the @xmath109 is certainly a goldstone mode .
@xmath112 is a pseudoscalar vertex and can therefore be expressed @xmath113.\ ] ] equation([avwti ] ) can now be used to derive a collection of chiral - limit , pointwise goldberger - treiman relations , important amongst which is the identity @xmath114 where @xmath115 is obtained in solving the chiral - limit gap equation .
it is now plain that if @xmath116 then @xmath117 .
this being true , then the homogeneous bethe - salpeter equation for the @xmath109 does not possess a massless solution in the chiral limit .
the converse is also true ; namely , the absence of such a solution requires eq.([caleb ] ) .
hence , eq.([caleb ] ) is a necessary and sufficient condition for the absence of a massless @xmath109 bound - state .
it is the chiral limit that is being discussed , in which case @xmath118 if , and only if , chiral symmetry is dynamically broken .
hence , the absence of a massless @xmath109 bound - state is only assured through the existence of an intimate connection between dcsb and an expectation value of the topological charge density .
a relationship between the mechanism underlying dcsb and the absence of a ninth goldstone boson was also discussed in ref.@xcite .
reference@xcite derives a range of corollaries , amongst which are mass formulae for neutral pseudoscalar mesons .
these led , e.g. , to a prediction of the manner by which the @xmath109 is split from the octet pseudoscalars by an amount that depends on qcd s topological susceptibility .
that is most easily illustrated by considering the @xmath89 limit , in which all current - quark masses assume the single value @xmath119 . in this case one
finds @xmath120 where @xmath121 is the meson s mass , @xmath122 with @xmath123 , and the residue of the @xmath109 bound - state pole in @xmath112 of eq.([au ] ) is @xmath124 plainly , the @xmath109 is split from the goldstone modes so long as @xmath125 .
it is argued @xcite that in qcd @xmath126 and it can be seen to follow from the gap equation , the homogeneous bse and eqs.([fpiacpres ] ) that @xmath127 consider now eq.([etapchiral ] ) in the form @xmath128 the first term is zero in the limit @xmath129 while the second remains finite . subsequently taking the chiral limit
, @xmath121 vanishes in the manner characteristic of all goldstone modes . before the chiral limit because the procedures do not commute @xcite . ]
these results are realised in the effective lagrangian of ref.@xcite in a fashion that is consistent with all the constraints of the anomalous ward identity .
this is _ not _ true of the so - called thooft determinant @xcite .
reference@xcite also presents an _ ansatz _ for the bethe - salpeter kernel that enables illustration of the implications of the mass formulae , in particular mixing between neutral pseudoscalars . despite its simplicity ,
the model is elucidative and phenomenologically efficacious ; e.g. , it predicts @xmath130@xmath109 mixing angles of @xmath131 , @xmath132@xmath130 angles of @xmath133 , and a strong neutron - proton mass difference of @xmath134 .
the use of dses to study meson phenomena is empowered by the existence noted above of a systematic , nonperturbative and symmetry - preserving truncation scheme @xcite .
it means that exact results , such as those outlined heretofore , can be illustrated and predictions for experiment made with readily quantifiable errors .
the renormalisation - group - improved rainbow - ladder truncation of the gap and bethe - salpeter equations introduced in refs.@xcite has been widely employed in this endeavour .
to exemplify that , in table[masses ] calculated results are reported for the masses of the lightest @xmath1 states @xcite .
it is true in general that the truncation is accurate for the @xmath135 and @xmath136 light - quark meson ground states . in these channels
it can be seen algebraically that contributions beyond rainbow - ladder largely cancel between themselves owing to eq.([avwti ] ) @xcite .
the remaining columns in the table deserve special attention because they show clearly the path toward improvement .
l||l|l||l||l|l||l|l|l ' '' '' @xmath137 & @xmath135 & @xmath136 & @xmath138 & @xmath139 & @xmath140 & @xmath141 & @xmath142 & @xmath143 + @xmath144 & 139 & 740 & 670 & 830 & 900 & 860 & 1000 & 1040 + @xmath145 & 695 & 1065 & 1080 & 1165 & 1240 & 1170 & 1310 & 1385 + terms beyond the rainbow - ladder truncation are known to add constructively in the @xmath138 channel @xcite .
hence the leading order truncation is _ a priori _ not expected to provide a good approximation .
further understanding is provided by an exploration of the contribution from two - pion intermediate states to the mass and width of this lowest - mass scalar .
a rudimentary analysis shows that a realistic description is attainable therewith @xcite ; viz .
, it gives a pole position @xmath146gev .
this is not the end of the scalar story but it is a sensible path to follow , in particular because a qcd - level mechanism is precisely specified .
something analogous should be considered in connection with the @xmath145 scalar state listed in table [ masses ] .
it is notable that in rainbow - ladder truncation it has been shown that at least up to the c - quark mass the ordering of meson masses is @xmath147 , where @xmath148 denotes the ground state and first radial excitation , respectively @xcite .
compared with experiment , the masses of the axial - vector mesons @xmath149 are poorly described by the rainbow - ladder truncation : @xmath150mev of repulsion is missing from the kernel .
a cruder model does better @xcite .
the latter studies and a more recent analysis @xcite indicate that at least part of the defect owes to the absence of spin - flip contributions at leading - order .
such contributions appear at all higher orders and are enhanced by the strongly dressed quark mass function .
this is one of the ways that the meson spectrum can be used to probe the long - range part of the light - quark interaction and thereby to chart the nonperturbative behavior of qcd s @xmath57-function .
the last three columns describe systems with so - called exotic quantum numbers . of course , these states are exotic only in the context of the naive constituent quark model . in qcd
they correspond simply to interpolating fields with some gluon content and are easily accessible via the bse @xcite .
nonetheless , while the rainbow - ladder truncation binds in these channels , the shortcomings encountered in the @xmath151 channels are also evident here , for many of the same reasons . reliable predictions for the masses of such states will only be obtained once improved kernels are developed . at the very least
, one must have dependable predictions for axial - vector masses before drawing any conclusions about the so - called exotics . since all these states lie within a domain
on which dcsb is very relevant , a framework with a veracious description of that emergent phenomenon is essential .
one might pose the question of whether , in the context of bound - state studies in which model assumptions are made regarding the nature of the long - range interaction between light quarks , anything is gained by working solely with schwinger functions .
this means , in part , constraining oneself to work only with information obtained from the dses at spacelike momenta .
according to a recent study @xcite the answer is no .
it analysed the capacity of schwinger functions to yield information about bound states , and established that for the ground state in a given channel the mass and residue are accessible via rudimentary methods .
however , simple methods can not provide dependable information about more massive states in a given channel .
indeed , there is no easy way to extract such information .
an approach based on a correlator matrix can be successful but only if the operators are carefully constructed so as to have large overlap with states of interest in a given channel , and statistical and systematic errors can be made small ; viz . , @xmath152% . while it is possible in principle to satisfy these constraints
, doing so is labor intensive and time consuming .
that is only justified in the absence of model - dependence .
partly because attention to this sector has only recently increased , the current level of expertise in studying baryons is roughly the same as it was with mesons more than ten years ago ; viz . , model building and phenomenology .
we are a little ahead of that game , however , because much has been learnt in meson applications ; e.g. , as indicated above , a veracious understanding of the structure of dressed - quarks and -gluons has been acquired . in quantum field theory
a nucleon appears as a pole in a six - point quark green function .
the pole s residue is proportional to the nucleon s faddeev amplitude , which is obtained from a poincar covariant faddeev equation that adds - up all possible quantum field theoretical exchanges and interactions that can take place between three dressed - quarks .
this is important because modern , high - luminosity experimental facilities employ large momentum transfer reactions ; viz .
, @xmath153 where @xmath154 is the nucleon s mass , and hence a veracious understanding of contemporary data requires a poincar covariant description of the nucleon .
a tractable truncation of the faddeev equation is based @xcite on the observation that an interaction which describes mesons also generates diquark correlations in the colour-@xmath155 channel @xcite .
the dominant correlations for ground state octet and decuplet baryons are @xmath156 and @xmath157 diquarks because , e.g. : the associated mass - scales are smaller than the baryons masses @xcite , namely ( in gev ) @xmath158_{0^+ } } = 0.7 - 0.8 \,,\ ; m_{(uu)_{1^+}}=m_{(ud)_{1^+}}=m_{(dd)_{1^+}}=0.9 - 1.0\,;\ ] ] and the electromagnetic size of these correlations , while larger than the pion , is less than that of the proton @xcite @xmath159_{0^+ } } \approx 0.7\,{\rm fm}\,,\ ; r_{(ud)_{1^+ } } \sim 0.8\,{\rm fm}\,.\ ] ] the last result is an estimate based on the @xmath160-meson/@xmath161-meson radius - ratio @xcite .
the faddeev equation s kernel is completed by specifying that the quarks are dressed , with two of the three dressed - quarks correlated always as a colour-@xmath155 diquark . binding is then effected by the iterated exchange of roles between the bystander and diquark - participant quarks .
a ward - takahashi - identity - preserving electromagnetic current for the baryon thus constituted is subsequently derived @xcite .
it depends on the electromagnetic properties of the axial - vector diquark correlation .
a study of the nucleon s mass and the effect on this of a pseudoscalar meson cloud are detailed in @xcite .
lessons learnt were employed in a series of studies of nucleon properties , including form factors @xcite .
the calculated ratio @xmath162 passes through zero at @xmath163gev@xmath35 @xcite . for the neutron , in the neighbourhood of @xmath164 , @xmath165 , where @xmath166 is the neutron s electric radius @xcite .
the evolution of @xmath162 and @xmath167 on @xmath168gev@xmath35 are both primarily determined by the quark - core of the nucleon .
while the proton ratio decreases uniformly on this domain @xcite , the neutron ratio increases steadily until @xmath169gev@xmath35 @xcite .
a form factor ratio motivated by ideas from perturbative qcd @xcite is depicted in fig.[nf2f1 ] .
the parameter @xmath61 is interpreted as a mass - scale that defines the upper - bound on the domain of so - called soft momenta in the perturbative analysis .
a plausible value for such a quantity is @xmath170 @xcite.gev corresponds to a length - scale @xmath171fm .
it is not credible that perturbative qcd is applicable at ranges greater than the proton s radius .
a reasonable value is @xmath172gev , leading to @xmath173fm . ]
it is interesting that the model of ref.@xcite yields neutron and proton ratios which cross at @xmath174gev@xmath35 .
the model s prediction for truly asymptotic momenta is currently being explored .
it is plausible that nature s fundamental `` constants '' might actually exhibit spatial and temporal variation @xcite .
this motivates an exploration of the current - quark - mass - dependence of the nucleon magnetic moments , which complements work on hadron @xmath76-terms @xcite .
preliminary results for the quark - core contribution to this variation can be expressed through the following ratios evaluated at the physical current - quark mass @xcite : @xmath175 it is likely that pseudoscalar meson contributions will increase these values by a factor of @xmath176 @xcite .
the framework can naturally be applied to calculate weak and strong form factors of the nucleon .
preliminary studies of this type are reported in refs.@xcite .
such form factors are sensitive to different aspects of quark - nuclear physics and should prove useful , e.g. , in constraining coupled - channel models for medium - energy production reactions on the nucleon .
this is important to the search for the so - called missing nucleon resonances ) , which are predicted by the @xmath177 constituent quark model ( @xmath178 ) , are not seen in the baryon spectrum that is obtained from amplitude analyses of @xmath179 elastic scattering . at this time more than half of
the low - lying states predicted by @xmath178 are missing @xcite . ] and the related problem of identifying exotic baryons .
of significant interest is the distribution of an hadron s _ spin _ over the quark constituents and their angular momentum . in a poincar covariant approach that can be calculated in any frame .
the rest frame is physically most natural .
the pion was considered in ref.@xcite and although @xmath180 the dressed - quarks carry significant orbital angular momentum .
a mixing angle of @xmath181 can be attributed in the rest frame to the @xmath182 components of the pion s poincar covariant bethe - salpeter wave function .
the answer is more complicated for the spin-@xmath183 nucleon . in the truncation described above , a nucleon s faddeev wave - function
is expressed through eight scalar functions : no more are needed , no number fewer is complete .
two are associated with the @xmath156 diquark correlation : @xmath184 , and six with the @xmath157 correlation : @xmath185 . in the rest frame in this basis one can derive the following `` good '' angular momentum and spin assignments , which add vectorially to give a @xmath186 nucleon : ) herein describes the correct assignments . ]
@xmath187\hline { \cal s}_1\,,{\cal a}_{2}\ , , { \cal b}_1 & { \cal s}_2\ , , { \cal a}_1\ , , { \cal b}_2\ , , & { \cal c}_2 & { \cal c}_1 \end{array}\ ; , \\ & & \begin{array}{cclcclcclccl } b_1 & = & \frac{1}{3 } a_3 + \frac{2}{3 } a_5\ , , & b_2 & = & \frac{1}{3 } a_4 + \frac{2}{3 } a_6\ , , & c_1 & = & a_3 - a_5\ , , & c_2 & = & a_4 - a_6 \ , .
\end{array } \label{ls}\end{aligned}\ ] ] these assignments are straightforward to demonstrate and understand ; e.g , in the rest frame of a relativistic constituent quark model the @xmath184 terms correspond , respectively , to the upper and lower components of the nucleon s spinor . to exhibit the importance of the various @xmath188-@xmath58 correlations within the nucleon s faddeev wave - function , ref.@xcite reported the breakdown of contributions to the nucleon s canonical normalisation : indicates the integrated contribution associated with @xmath189 .
the entries are reweighted such that the sum of the squares of the entries equals one .
positions without an entry are zero to two decimal places . ]
@xmath190 to illustrate how eq.([lsresult ] ) should be understood , note that the largest single entry is associated with @xmath191 , which represents the quark outside the scalar diquark correlation carrying all the nucleon s spin .
that is the @xmath100-quark in the proton .
however , it is noteworthy that a contribution of similar magnitude is associated with the axial - vector diquark correlations , which express mixing between @xmath192- and @xmath193-waves ; viz . , @xmath194 .
with @xmath195 all quark spins are aligned with that of the nucleon and the unit of angular momentum is opposed : @xmath196 , while with @xmath197 all quark spins are opposed and the two units of angular momentum are aligned : @xmath198 . this contribution is more important than those associated with @xmath199 ; namely , scalar diquark terms with the bystander quark s spin antiparallel .
finally , for the present , in this context one single number is perhaps most telling : the contribution to the normalisation from @xmath200 terms is only 37% of the total .
the basic problem of hadron physics is to solve qcd . in order to achieve this goal a joint effort from experiment and theory
is required .
our community now has a range of major facilities that are accumulating data , of unprecedented accuracy and precision , which pose important challenges for theory .
it is the feedback between experiment and theory that leads most rapidly to progress in understanding . the opportunities for hadron physics will grow because upgraded and new facilities will appear on a five - to - ten - year time - scale .
this volume provides an excellent illustration of the splendid experimental accomplishments of recent times , the achievements of theory in explaining and interpreting the observations , and some of the urgent open problems .
two of the things that stand out are the diversity of theoretical approaches and the interplay between them that is necessary to draw valid conclusions from contemporary data .
one need not think too hard in order to list a number of the key open problems .
it is essential to understand the origin and the nature of confinement .
asymptotic coloured states have not hitherto been observed , but is it a cardinal fact that they can not ?
as described herein , confinement is related to the analytic structure of elementary n - point functions in qcd and especially to their properties at infrared momenta .
is it in the essence of non - abelian gauge theories that they possess an hilbert space of physical states that is reconstructed from only a small subset of all the theory s schwinger functions ? in progressing toward an answer to the question of confinement
it will be necessary to map out the infrared behaviour of qcd s @xmath57-function .
the upgraded and future facilities will provide data which will guide that process .
however , to make full use of that data , it will be necessary to have reliable poincar covariant theoretical tools which enable the study of hadrons in the mass range @xmath201-@xmath202gev .
on this domain confinement and dynamical chiral symmetry breaking are both germane .
it is known that dynamical chiral symmetry breaking ( dcsb ) ; namely , the generation of mass _ from nothing _ , does take place in qcd .
however , the origin of the interaction strength at infrared momenta that guarantees dcsb through the gap equation is unknown .
this ties confinement to dcsb .
further to this connection , at nonzero temperature and chemical potential are the deconfinement and chiral symmetry restoration transitions simultaneous ?
if so , why , under what conditions , and also , perhaps , in which class of theories ?
it is important to understand the relationship between parton properties on the light - front and the rest frame structure of hadrons .
this is a problem because , e.g. , dcsb , an established keystone of low - energy qcd , has not been realised in the light - front formulation .
the obstacle is the constraint @xmath203 for massive quanta on the light front @xcite .
it is therefore impossible to make zero momentum fock states that contain particles and hence the vacuum is trivial . only the zero modes of light - front quantisation can dress the ground state but little progress has been made with understanding just how that might occur .
furthermore , parton distribution functions must be calculated in order to comprehend their content .
parametrisation is insufficient .
it would be very interesting to know , e.g. , how , if at all , the distribution functions of a goldstone mode differ from those of other hadrons . calculation @xcite and experiment @xcite together are needed to address that issue . a last question in this brief survey , but certainly not the least ,
how do the properties of hadrons change within nuclei , and within very dense systems ? absent an answer to this , one is not addressing the subject of nuclear physics .
the experimental archetype for this problem is the emc effect .
experiments have shown that in - nucleus parton distribution functions can not simply be obtained by adding together the distributions within the constituent nucleons .
a satisfactory understanding of this effect is still lacking , almost twenty - five years after its discovery . within
this theme one also asks for information on : the equation of state for dense matter ; the properties of the hadron to quark - gluon phase transition ; and the evolution between and nature of the states of matter within compact astrophysical objects .
plainly , in addressing this question one is embarking on a far reaching quest .
there is ample evidence of the constructive feedback between theoretical approaches .
for example , dyson - schwinger equation studies and effective field theory inform calculations within lattice - regularised qcd , and this interplay will become increasingly powerful as the parameters of lattice - qcd simulations come closer to the physical domain .
notwithstanding the investment in and progress that has been made with lattice - regularised qcd , for example , there is a pressing present and continuing need for model building and application in hadron physics .
no single so - called _
ab - initio _ approach is currently applicable to all the phenomena attributed to qcd at all the length - scales that can be explored .
therefore , understanding and synthesising the wealth of extant and forthcoming data requires the continued development of models .
such tools answer an immediate need and provide a ready means to develop intuition about a complex system .
naturally , a reasonable model should obey all relevant constraints that _ ab - initio _ approaches and symmetries apply .
that being so , it can then provide a rational and flexible guide in connection with our rapidly changing experimental environment .
continued investment in each of the nonperturbative tools that might provide insight into qcd is also necessary .
perturbative qcd is understood but its application begs a key question : at just which scale does its employment become valid ?
the answer can only be attained nonperturbatively .
a number of experiments are described as exhibiting precocious scaling ; i.e. , they are associated with cross - sections that appear to evolve with momentum according to perturbative qcd expectations .
however , only a comprehensive nonperturbative analysis can explain whether that is the result of an accidental cancellation over the often small momentum domain accessible to the experiment or a clear signal of perturbative qcd .
partial answers to this and related questions are possible through internally consistent model calculations that can unify many observables , e.g. , ref.@xcite .
but a complete answer will require truly nonperturbative calculations of the scale - dependence of observables within a framework that is able to exhibit the perturbative qcd limit , e.g. , ref.@xcite .
this contribution has outlined the utilisation of dses in studying some of the many facets of the basic problem of hadron physics .
the dses provide a natural framework within which light - quark confinement can be defined , and the origin and consequences of dcsb explored . to build an understanding of confinement , it is essential to work toward an accurate map of the interaction between light - quarks . among the rewards
are a precise connection between confinement and dcsb , an accounting of the distribution of mass within hadrons , and a realistic picture of hybrids and exotics
. it can not be too strongly emphasised that dcsb is remarkably effective at generating mass . for light - quarks
it is far more important than the higgs mechanism .
it is understood via qcd s gap equation , which delivers a quark mass function with a momentum - dependence that connects the perturbative domain with the nonperturbative , constituent - quark domain .
the existence of a sensible truncation scheme enables the proof of exact results using the dses .
the scheme is also tractable , and hence the results can be illustrated and predictions made for observables .
the consequent opportunities for rapid feedback between experiment and theory provide openings through which an intuitive understanding of numerous nonperturbative strong interaction phenomena can be reached via the dses .
i acknowledge valuable conversations with b. el - bennich , i. c. clot , t. klhn , a. krassnigg and r. d. young .
i am grateful to the organisers , and to the staff at the _ ettore majorana foundation and centre for scientific culture _ , for their hospitality and support during the meeting .
this work was supported by the department of energy , office of nuclear physics , contract no .
de - ac02 - 06ch11357 ; and benefited from the facilities of the anl computing resource center .
f. t. hawes , c. d. roberts and a. g. williams , phys .
d * 49 * , 4683 ( 1994 ) .
m. s. bhagwat , _
et al_. , _ few body syst .
_ * 40 * , 209 ( 2007 ) .
m. stingl , _ phys .
_ d34 ( 1986 ) 3863 [ erratum - ibid .
d * 36 * ( 1987 ) 651 ] a. bender , d. blaschke , yu .
l. kalinovsky and c. d. roberts , _ phys .
_ 77 ( 1996 ) 3724 a. bender ,
_ et al_. , _ phys .
_ b431 ( 1998 ) 263 c. d. roberts and a. g. williams , _ prog .
_ 33 ( 1994 )
477 h. j. munczek and a. m. nemirovsky , _ phys . rev .
_ d28 ( 1983 ) 181 f. t. hawes , p. maris and c. d. roberts , _ phys .
_ b440 , 353 ( 1998 ) m. s. bhagwat , m. a. pichowsky , c. d. roberts and p. c. tandy , _ phys .
_ c68 ( 2003 ) 015203 m. s. bhagwat , _
et al_. , _ phys .
_ * c70 * , 035205 ( 2004 ) .
m. s. bhagwat and p. c. tandy , _ phys .
_ d70 ( 2004 ) 094039 h. h. matevosyan , a. w. thomas and p. c. tandy , _ phys .
_ c75 ( 2007 ) 045201 j. i. skullerud , _ et al_. , _ jhep _ 0304 ( 2003 ) 047 c. j. burden and c. d. roberts , _ phys .
_ d47 ( 1993 ) 5581 z. h. dong , h. j. munczek and c. d. roberts , _ phys .
_ b333 ( 1994 ) 536 a. bashir and m. r. pennington , _ phys .
_ d50 ( 1994 ) 7679 j. s. ball and t. w. chiu , _ phys .
_ d22 ( 1980 ) 2542 c. j. burden , c. d. roberts and a. g. williams , _ phys .
lett . _ b285 ( 1992 ) 347 d. v. widder , _ bull .
_ 40 ( 1934 ) 321 l. von smekal , a. hauck and r. alkofer , _ annals phys .
_ 267 ( 1998 ) 1 [ _ erratum - ibid . _ 269 ( 1998 ) 182 ] r. alkofer , w. detmold , c. s. fischer and p. maris , _ phys
_ d70 ( 2004 ) 014014 m. gopfert and g. mack , _ commun .
_ 82 ( 1981 ) 545 a. bashir and a. raya , `` truncated schwinger - dyson equations and gauge covariance in qed3 , '' _ few body syst .
_ on - line , doi 10.1007/s00601 - 007 - 0177 - 3 p. pascual and r. tarrach , _ qcd : renormalization for the practitioner _ , springer lecture notes in physics , 194 ( springer verlag , berlin / heidelberg , 1984 ) .
available on - line at http://www.springerlink.com/content/t76570333v7g c. d. roberts , _ nucl .
phys . _ a605 ( 1996 ) 475 c. d. roberts and s. m. schmidt , _ prog .
_ 45 ( 2000 ) s1 r. alkofer and l. von smekal , _ phys .
_ 353 ( 2001 ) 281 p. maris and c. d. roberts , _ int
. j. mod .
_ e12 ( 2003 ) 297 c. d. roberts , m. s. bhagwat , s. v. wright and a. hll , _ eur .
j. special topics _ 140 ( 2007 ) 53 p. maris , c. d. roberts and p. c. tandy , _ phys .
_ b420 ( 1998 ) 267 m. s. bhagwat and p. c. tandy , _ aip conf .
* 842 * ( 2006 ) 225 p. o. bowman , _
et al_. , _ phys .
_ d71 ( 2005 ) 054507 m. s. bhagwat , i. c. clot and c. d. roberts , `` covariance , dynamics and symmetries , and hadron form factors , '' arxiv:0710.2059 [ nucl - th ]
. k. langfeld , _
et al_. , _ phys .
_ c67 ( 2003 ) 065206 l. chang , _
et al_. , _ phys .
rev . _ c75 ( 2007 ) 015201 p.
maris and c. d. roberts , _ phys .
_ c56 ( 1997 ) 3369 m. s. bhagwat , _
et al_. , _ eur .
j. _ a31 ( 2007 ) 630 a. hll , p. maris , c. d. roberts and s. v. wright , _ nucl .
_ 161 ( 2006 ) 87 h. j. munczek , _ phys
_ d52 ( 1995 ) 4736 a. bender , c. d. roberts and l. von smekal , _ phys .
lett . _ b380 ( 1996 ) 7 a. hll , a. krassnigg and c. d. roberts , _ phys .
_ c70 ( 2004 ) 042203(r ) a. hll , _
et al_. , _ phys .
_ c71 ( 2005 ) 065204 c. mcneile and c. michael [ ukqcd collaboration ] , _ phys .
lett . _ b642 ( 2006 ) 244 m. a. ivanov , yu .
l. kalinovsky , p. maris and c. d. roberts , _ phys .
lett . _ b416 ( 1998 ) 29 m. a. ivanov , yu .
l. kalinovsky and c. d. roberts , _ phys .
_ d60 ( 1999 ) 034018 p. maris and p. c. tandy , _ phys .
_ c60 ( 1999 ) 055214 p. maris and p. c. tandy , _ phys .
_ c62 ( 2000 ) 055204 m. a. ivanov , j. g. korner , s. g. kovalenko and c. d. roberts , _ phys .
_ d76 ( 2007 ) 034018 m. s. bhagwat , _
et al_. , _ phys .
_ c76 ( 2007 ) 045203 a. e. dorokhov and w. broniowski , _ eur .
j. _ c32 ( 2003 ) 79 j. b. kogut and l. susskind , _ phys .
_ d11 ( 1975 ) 3594 e. witten , _ nucl .
_ b156 ( 1979 ) 269 g. veneziano , _ nucl .
_ b159 ( 1979 ) 213 r. narayanan and h. neuberger , _ nucl .
_ b696 ( 2004 ) 107 p. di vecchia , _ phys .
_ b85 ( 1979 ) 357 r. j. crewther , _ phys .
_ b70 ( 1977 ) 349 r. j. crewther , _ acta phys .
austriaca suppl .
_ 19 ( 1978 ) 47 g. a. christos , _ phys .
rept . _ 116 ( 1984 ) 251 d. kekez and d. klabucar , _ phys .
_ d73 ( 2006 ) 036002 i. c. cloet , a. krassnigg and c. d. roberts , `` dynamics , symmetries and hadron properties , '' arxiv:0710.5746 [ nucl - th ] .
a. bender , w. detmold , c. d. roberts and a. w. thomas , _ phys .
_ c65 ( 2002 ) 065203 c. d. roberts , `` confinement , diquarks and goldstone s theorem , '' in _ proc .
2nd int .
conf . on quark confinement and the hadron spectrum _ ,
n. brambilla and g. m. prosperi ( world scientific , singapore , 1997 ) p. 224 .
a. krassnigg , c. d. roberts and s. v. wright , _ int .
j. mod .
_ a22 ( 2007 ) 424 c. j. burden , _
et al_. , _ phys .
_ c55 ( 1997 ) 2649 j. c. r. bloch , yu .
l. kalinovsky , c. d. roberts and s. m. schmidt , _ phys .
_ d60 ( 1999 ) 111502(r ) p. watson , w. cassing and p. c. tandy , _ few body syst .
_ 35 ( 2004 ) 129 ( 2004 ) c. j. burden and m. a. pichowsky , _ few body syst .
_ 32 ( 2002 ) 119 r. t. cahill , c. d. roberts and j. praschifka , _ austral
. j. phys . _
42 ( 1989 ) 129 r. t. cahill , c. d. roberts and j. praschifka , _ phys .
_ d36 ( 1987 ) 2804 p. maris , p. maris , _ few body syst .
_ 35 ( 2004 ) 117 m. s. bhagwat and p. maris , `` vector meson form factors and their quark - mass dependence , '' arxiv : nucl - th/0612069 . m. oettel , m. pichowsky and l. von smekal , _ eur . phys . j. _ a8 ( 2000 ) 251 m. b. hecht , _ et al_. , _ phys . rev . _ c65 ( 2002 ) 055204 r. alkofer , _
et al_. , _ few body syst .
_ 37 ( 2005 ) 1 a. hll , _ et al_. , _ nucl .
_ a755 ( 2005 ) 298 m. s. bhagwat , a. hll , a. krassnigg and c. d. roberts , _ nucl .
. _ a790 ( 2007 ) 10 a. v. belitsky , x. d. ji and f. yuan , _ phys . rev .
_ 91 ( 2003 ) 092003 j. p. uzan , _ rev .
_ 75 ( 2003 ) 403 v. v. flambaum , _
et al_. , _ few body syst .
_ 38 ( 2006 )
31 v. v. flambaum and r. b. wiringa , _ phys . rev .
_ c76 ( 2007 ) 054002 v. v. flambaum , d. b. leinweber , a. w. thomas and r. d. young , _ phys .
_ d69 ( 2004 ) 115006 j.c.r .
bloch , c.d .
roberts and s.m .
schmidt , _ phys .
_ c61 ( 2000 ) 065207 ( 2000 ) m. oettel , r. alkofer and l. von smekal , _ eur .
j. _ a8 ( 2000 ) 553 a. hll , c. d. roberts and s. v. wright , _ aip conf .
* 857 * ( 2006 ) 46 t. s. lee and l. c. smith , _
j. phys . _ g34 ( 2007 ) s83 m. oettel , _
et al_. , _ phys .
_ c58 ( 1998 ) 2459 s. j. brodsky and h. c. pauli , _ lect .
notes phys .
_ 396 ( 1991 ) 51 m. b. hecht , c. d. roberts and s. m. schmidt , _ phys . rev .
_ c63 ( 2001 ) 025213 k. wijesooriya , p. e. reimer and r. j. holt , _ phys .
rev . _ c 72 ( 2005 ) 065203 p.
maris and c. d. roberts , _ phys .
_ c58 ( 1998 ) 3659 | an overview of the theory and phenomenology of hadrons and qcd is provided from a dyson - schwinger equation viewpoint . following a discussion of the definition and realisation of light - quark confinement ,
the nonperturbative nature of the running mass in qcd and inferences from the gap equation relating to the radius of convergence for expansions of observables in the current - quark mass are described . some exact results for pseudoscalar mesons
are also highlighted , with details relating to the @xmath0 problem , and calculated masses of the lightest @xmath1 states are discussed .
studies of nucleon properties are recapitulated upon and illustrated : through a comparison of the @xmath2-weighted ratios of pauli and dirac form factors for the neutron and proton ; and a perspective on the contribution of quark orbital angular momentum to the spin of a nucleon at rest .
comments on prospects for the future of the study of quarks in hadrons and nuclei round out the contribution . |
astronomical observations indicate that many galaxies host a supermassive black hole ( smbh ) in their centre .
the inspirals of stellar - mass compact objects into such smbhs with mass @xmath0@xmath1 constitute one of the most important gravitational wave ( gw ) sources for the planned space - based gw observatory lisa .
the large parameter space of possible signals and their long duration makes detection using fully coherent matched filtering computationally intractable , so alternative search techniques are required .
one option is to use a semi - coherent matched filtering algorithm and a preliminary analysis of that approach @xcite suggests that the lisa emri detection rate will be dominated by inspirals of @xmath2 bhs onto @xmath3 smbhs , and could be as high as @xmath4 in 35 years within @xmath5 gpc . matched filtering algorithms are computationally very expensive and it is therefore valuable to scope out alternatives . in a previous paper @xcite , we described a strategy to detect gws from emris by accumulating the signal power in the time - frequency ( t - f ) domain .
the t - f power spectrum is produced by dividing the data into segments of equal duration and carrying out a fast fourier transform ( fft ) on each .
the length of the data segments affects the resolution of the spectrogram and must be chosen to allow the evolution of the source to be followed in the t - f plane .
this is a search parameter that should eventually be optimized , probably in a source specific way .
for now we choose @xmath6 week segments , as these provide appropriate resolution for typical emris
. the choice of segment duration does not significantly affect the computational requirements of the search , in contrast to the semi - coherent approach @xcite which is limited to @xmath7 week segments by computational cost .
the power spectrum is defined for each segment @xmath8 and frequency bin @xmath9 as , @xmath10 = \sum_{\alpha = i , ii } \left[\frac{2 ( h^{\alpha \ , i}_k)^{2}}{\sigma^2_{n^{\alpha}_k}}+4\frac{\mbox{re}[h^{\alpha \ , i}_k ( n^{\alpha \ , i}_k)^ { * } ] } { \sigma^2_{n^{\alpha}_k } } + \frac{2 ( n^{\alpha \ , i}_k)^{2}}{\sigma^2_{n^{\alpha}_k } } \right],\ ] ] [ power ] where @xmath11 denotes the fourier amplitude of the signal , @xmath12 is the expected variance of the noise component @xmath13 at frequency bin @xmath9 , characterized by @xmath14 , the strain spectral density of the noise , @xmath15 is the sampling cadence and @xmath16 is the fourier bin width .
we assume that the lisa noise components , @xmath13 , can be approximated by a stationary and gaussian distribution , with spectral density @xmath14 .
in addition to the instrumental noise , the lisa data stream will contain an unresolvable astrophysical foreground from galactic white dwarf ( wd ) binaries .
we include the white dwarfs as a noise contribution in @xmath17 in the usual way ( @xcite , equations ( 52 ) and ( 54 ) ) .
the instrumental noise spectral density is also taken from @xcite , equation ( 48 ) . including the wd foreground in this way
assumes it is isotropic on the sky , which is not a good assumption in this case .
for the same reason , the wd foreground will not be a stationary noise source on long timescales , but this is probably a reasonable approximation for few week long data segments . in the future
, our analysis will be repeated using a more accurate realization of the wd distribution ( e.g. , @xcite ) . generally speaking , emris are not detectable when their frequencies overlap with the wd confusion background , so the results in this paper should not change significantly
. however , the location in frequency of the confusion limit will be important , i.e. , which wd binaries are resolvable and which are not @xcite . in the low - frequency regime
, lisa can be regarded as consisting of two independent michelson detectors which are denoted i and ii @xcite .
we combine the responses from these two detectors to produce the power spectrum . in this first analysis
, we add the spectrograms from the i and ii responses before performing the search .
an alternative algorithm might involve searching each spectrogram separately and looking for coincidences .
such refinements will be investigated in the future . for the combined spectrogram
, we calculate the power `` density '' , @xmath18 , by computing the average power within a rectangular box centered at each point ( @xmath19 ) , @xmath20 where @xmath21 , @xmath22 are the lengths of the box edges in the time and frequency dimensions respectively and @xmath23 is the number of data points enclosed . to search for a possible signal ,
we vary the size of the box , and for each choice of @xmath21 and @xmath22 , @xmath24 say , we search for any points at which @xmath18 exceeds a threshold , @xmath25 .
a detection occurs when there is at least one point with @xmath26 for at least one box size . in the first algorithm
, we use only box sizes @xmath27 , for all possible integer choices of @xmath28 and @xmath29 .
the thresholds , @xmath25 , are set to ensure a small overall false alarm probability ( fap ) for the search .
this will be discussed in more detail later .
the optimal box size for a given source will be large enough to include much of the signal power but small enough to avoid too much noise contribution .
the optimum will be source specific due to the wide variation in emri waveforms .
the inspiral of a @xmath30white dwarf occurs much more slowly than that of a @xmath31black hole , so in the first case , the optimal box size is likely to be longer in the time dimension .
gws from an inspiral into a rapidly spinning black hole or from a highly eccentric inspiral orbit are characterized by many frequency harmonics , often close together . in that case
, a box that is wider in the frequency dimension may perform well . in designing a search
, a balance must therefore be struck between having sensitivity to a range of sources and increasing the reach of the search for a specific source . in section [ noisestats ]
we discuss the statistics of this search in the absence of a signal . in section [ threshchoice ]
we discuss how to assign thresholds for the different box sizes and in section [ algperf ] we illustrate the performance of the algorithm on a set of trial waveforms . we discuss some remaining issues and possible ways to develop time - frequency techniques in section [ conc ] .
if no signal is present , then the power @xmath32 at each point in the unbinned plane is distributed as a @xmath33 with 4 degrees of freedom .
there are four degrees of freedom since there are two independent michelson responses and the real and imaginary parts of @xmath34 are independent for each
. over many noise realizations , the total binned power , @xmath35 , at a specific point in the spectrogram will be distributed like a @xmath33 with @xmath36 degrees of freedom .
the probability that @xmath35 exceeds a threshold @xmath25 is therefore @xmath37 ( where @xmath38 is the cumulative distribution function for a @xmath33 with @xmath39 degrees of freedom ) .
if @xmath40 independent samples are drawn from such a distribution , the probability that the maximum , @xmath41 , is less than a specified value , @xmath42 , is @xmath43^q$ ] . however , in the binned t - f plane , @xmath35 is not independent at nearby points since neighbouring bins overlap .
there are at least @xmath44 independent samples in the plane , but less than @xmath45 ( where @xmath45 is the total number of pixels in the unbinned spectrogram ) . in @xcite
, we used the estimate that there are @xmath46 independent points . in practice
, the distribution of the maximum is not analytically tractable , but it can be computed numerically using monte carlo simulations . from a monte carlo simulation of a quarter of a million noise realizations , we have computed the distribution of the maximum power in the absence of a signal for each box size used .
these distributions are much more accurate than the approximations used in @xcite .
the distributions were computed assuming we had @xmath47s ( @xmath48 years ) of lisa data , sampled at a cadence of @xmath49s and divided this into @xmath50s ( @xmath6 week ) sections .
this gives a time - frequency spectrogram with @xmath51 points in time and @xmath52 points in frequency and hence @xmath53 possible box sizes of the form @xmath54 , @xmath55 .
for each noise realization , we computed the time - frequency spectrogram of the lisa mission and then calculated the maximum power in each binned spectrogram .
figure [ sampledist ] illustrates one typical distribution , with @xmath56 and @xmath57 .
it also shows the approximate theoretical distribution described above , under the assumption that there are @xmath58 , @xmath44 or @xmath46 independent samples in the t - f plane .
as expected , the true distribution lies between the theoretical distributions for @xmath45 and @xmath44 independent points .
our first approximation of @xmath46 independent points @xcite is not particularly accurate . assuming @xmath59 independent points gives a better approximation to the distribution in this case , although the shape is clearly different
this number is surprisingly large , indicating that boxes with significant overlap are still effectively independent .
this suggests that the maximum is usually determined by a few unusually bright pixels , rather than sizable regions of high power .
the effective number of independent boxes is box size specific , but always lies between @xmath44 and @xmath45 . assigning a threshold , @xmath25 , for a given box size amounts to fixing the false alarm probability for that box size ,
fap@xmath60 ( i.e. , the probability that this threshold is exceeded by pure noise ) .
if the binned distributions were independent , the overall false alarm probability for the search , fap , would be fap = 1 - @xmath61 .
but as this is not the case , we must make use of the monte carlo simulation results again . from the numerical distributions
we can compute a threshold for each box size that sets the fap@xmath60 to a specified value .
we then use the monte carlo data and count the fraction of pure noise realizations in which the maximum power exceeds the threshold for at least one box size .
this is the overall search fap .
since the t - f plane encompasses the whole of the lisa data stream , the fap is the fraction of lisa missions in which we would expect to have a false alarm . in figure [ searchfap ] , the search fap is shown as a function of the ( equal ) fap@xmath60 assigned to each box size . for small fap@xmath60 , fap@xmath62fap@xmath60 , indicating that effectively @xmath63 of the @xmath64 box sizes used are independent .
at a fixed search fap , a threshold choice corresponds to dividing up the fap among the different box sizes .
if we have no prior knowledge of the signal , the natural way to do this is to treat all the box sizes equally , i.e. , set fap@xmath65 constant .
this can be done accurately by determining the thresholds from the numerical distributions , as in the previous section . in @xcite
we set thresholds to give uniform fap@xmath60 s under the theoretical approximation that @xmath66 was distributed as the maximum of @xmath67 independent samples from a @xmath68 distribution . using that approach ,
the true fap@xmath60 s are not equal , but we determined the true fap of the search for each threshold choice using a small monte carlo simulation .
if we are searching for a particular source , the threshold choice can be tuned . to illustrate this
, we use the ` typical ' emri described in @xcite ( source `` a '' ) .
this is the last three years of the inspiral of a @xmath31 black hole into a @xmath69 smbh of spin @xmath70 m , with orbital eccentricity @xmath71 at the start of the observation and inclination @xmath72 .
the frequency spread , time spread and snr of our target waveform in principle allow us to guess theoretically which box size will be optimal for detecting it .
however , the presence of noise means that the source will not always be detected in the same box size and indeed it will not always be the same part of the inspiral that is detected ( different box sizes are optimal for different stages of the inspiral ) . to illustrate this , we ran a monte carlo simulation of the search over several hundred noise realizations with source a present in the data , and recorded every time a threshold was exceeded , and for which box size this occurred .
the thresholds were set to give equal fap@xmath60 to all box sizes .
when this procedure was repeated without the signal , the number of detections was roughly independent of the box size , as we would expect . in figure [ whichbin ]
we show the fraction of ` detections ' ( i.e. , a threshold being exceeded ) as a function of the box size label .
it is clear that there are preferred box sizes for detecting this source , but more than one . for a given noise realization ,
the source might be detected ( i.e. , exceed the box threshold @xmath25 ) with more than one box size , but these box sizes do vary from realization to realization .
thus , a targeted search for source a should still use a number of box sizes . in figure
[ compthresh ] we compare the performance of the algorithm at detecting source a under these various choices for the thresholds .
we show the receiver operator characteristic ( roc ) curve , which plots the detection rate versus the overall search fap .
the detection rate was estimated by adding the signal waveform to a sequence of noise realizations , and running the search algorithm .
we used @xmath73 noise realizations , giving an accuracy in the detection rate of @xmath74 .
the theoretical threshold assignment used in @xcite gives performance that is comparable with the true ` equal fap@xmath60 ' threshold assignment . a targeted search , which uses only the eleven box sizes which have detection rates greater than @xmath75 in figure [ whichbin ] , and assigns equal fap@xmath60 to each , performs significantly better than the blind search
however , if we use the set of box sizes tuned for source a to detect other sources , the performance can be worse . for example , when searching for the source `` s '' ( described below ) at a distance of @xmath76 gpc and at a search fap of @xmath77 , the detection rate using thresholds tuned for source a is @xmath78 compared to @xmath79 in the untargeted search .
the search can also be tuned for source s , and a final data analysis scheme might involve using a targeted search for one type of inspiral followed by a targeted search for another type and so on .
however , using multiple tuned searches increases the overall false alarm probability and eventually will do no better than a blind search . without a much better knowledge of the likely parameters of emri sources in the lisa data , it is probably best to treat all box sizes equally . in the next section , we will use thresholds that give equal fap@xmath60 to all box sizes .
to assess the algorithm s performance , we computed roc curves for a sequence of inspirals at a number of distances .
the inspiral waveforms were computed using the ` numerical kludge ' approximation @xcite , with detector modulations added using the low frequency approximation to the lisa response @xcite .
the detector response can be treated more accurately @xcite but the low frequency response is sufficient for this analysis .
our ` typical ' source , `` a '' , has the parameters specified above and was placed at a sky position @xmath80 , @xmath81 , with orientation chosen as @xmath82 , @xmath83 .
the angles @xmath84 and @xmath85 are the ecliptic colatitude and azimuth of the source on the sky , while the angles @xmath86 and @xmath87 are the ecliptic colatitude and azimuth to which the spin of the smbh would point in the solar system frame .
the other trial waveforms were obtained by changing one or two of the inspiral parameters .
the waveform parameters for the inspirals considered are listed in table [ waveparams ] . for each inspiral , the initial semi - latus rectum , @xmath88 , of the orbit
was chosen such that the plunge would occur a few days after the end of the lisa observation , i.e. , in each case we are focusing on approximately the last three years of inspiral .
table [ waveparams ] also lists the approximate signal to noise ratio that would be achieved in a fully coherent matched filtering search , using both michelson data streams ( i and ii ) , if the source was at a distance of @xmath89 gpc . for each trial waveform
we performed a monte carlo simulation , with the source at distances of @xmath90 , @xmath76 , @xmath91 , @xmath92 and @xmath93 gpc .
source `` a '' was additionally placed at distances of @xmath89 , @xmath94 , @xmath95 and @xmath96 gpc . in figure [ rocsourcea ]
we show the roc for detection of source `` a '' at each distance , in figure [ rocallsource2gpc ] we show the rocs for detection of all the sources at a distance of @xmath92 gpc and in table [ detectrates ] , we summarize the detection rates for each source at each distance , if the thresholds are set to give an overall search false alarm probability of @xmath97 . since this is the false alarm rate per lisa mission , we could afford to increase this number in the final search , up to @xmath98 or more
. however , for the present we use @xmath77 in order to compare the performance of this technique directly to that of the semi - coherent matched filtering search @xcite . in the latter search ,
a matched filtering snr of @xmath99 is required for detection when the overall search fap is set at @xmath77 .
no roc curves for the semi - coherent search are presently available , but the detection rate is estimated to be @xmath100 at threshold . with an overall search false alarm probability of @xmath77 , we see from figure [ rocsourcea ] that source a is easily detectable out to @xmath92 gpc .
the detection rate at greater distances decreases quite rapidly , although it is still @xmath101 at @xmath95 gpc .
the absolute limit of the search for source a is approximately @xmath93 gpc , at which point we do no better than a random search .
this compares quite favourably with the performance of the semi - coherent matched filtering search @xcite , which in principle can detect this source out to @xmath102 gpc in a euclidean universe .
in fact , in @xcite , the reach of the search was limited at a redshift @xmath103 , or @xmath104 gpc , by astrophysical considerations , since there is considerable uncertainty about the formation of emri sources at higher redshift .
varying the parameters of the inspiral has a noticeable effect on the detectability .
the white dwarf inspirals ( sources d , e and f ) are essentially not detectable ( they would be detectable very close by , but we do not expect any such sources closer than @xmath105 gpc ) . however , this is because these sources are weak and the snr is so low that they would not even be detected by the semi - coherent technique .
the intermediate mass black hole inspiral ( source g ) can be detected to great distances since it is very loud , but it is presently unclear whether any such sources are likely to exist @xcite . the inspirals into a lower ( source b ) or higher ( source c ) mass smbh are less detectable , the former becoming marginally detectable at @xmath92 gpc and the latter becoming marginally detectable at about @xmath95 gpc .
this is a consequence of the shape of the lisa noise curve . using the semi - coherent technique these sources are similarly less detectable than source a , although they can still be detected out to @xmath104 gpc . varying the eccentricity ( sources k
n ) has a moderate effect on detectability lower eccentricity sources can be detected somewhat further away .
this is in contrast to the semi - coherent technique , for which eccentricity has little effect on detectability .
low eccentricity inspirals have power spread over fewer frequency components , and so the power is more likely to be concentrated in certain regions in the t - f plane , making it easier for this algorithm to detect them . increasing the central smbh spin ( sources h
j ) appears to increase the detectability .
inspirals into highly spinning black holes will be at higher frequencies ( and thus over the floor of the instrumental noise ) and will plunge closer to the central smbh .
both of these properties increase the snr of the source and therefore are likely to increase the detectability by either this t - f method or the semi - coherent method , although this has not been investigated for the latter technique . increasing the inclination of the orbit with respect to the equatorial plane of the smbh ( sources o t ) also affects detectability , with lower inclination orbits being more detectable .
the inclination is defined @xcite such that an inclination of @xmath106 is a prograde equatorial orbit , while an inclination of @xmath107 is a retrograde equatorial orbit .
the increased detectability at low inclination is again largely due to the fact that such orbits plunge closer to the smbh and thus have higher snr .
finally , changing the extrinsic parameters ( source position on the sky and inclination of the smbh spin with respect to our line of sight ) also affects detectability ( sources extrinsa extrinsf ) . by ensuring that the source is in a sky position to which lisa is very sensitive , and that the black hole spin is oriented favourably with respect to the detector
, the detectability can be increased .
similarly , placing the source unfavourably can make detection marginal at @xmath92 gpc .
however , the extrinsic parameters chosen for source a seem to represent an ` average ' response performance . .parameters and
signal to noise ratios at @xmath89 gpc for trial waveforms .
unspecified parameters are the same as source `` a '' , as given in the text . [
cols="^,<,^,^",options="header " , ]
we have presented a thorough analysis of the time - frequency method to detect emris with lisa that was originally proposed in @xcite .
we have accurately computed the statistics of this search and examined its performance on a wide range of possible emri signals .
we find that this algorithm is able to detect many different emri events out to distances of @xmath108 gpc , depending on the source parameters . in an untargeted search
, a typical source can be detected at @xmath92 gpc with a detection rate of @xmath109 at a search false alarm probability of @xmath77 .
lower eccentricity sources , which have less frequency spreading , can be detected as far away as @xmath93 gpc with detection rates of @xmath79 at the same overall fap .
the reach of the search can be extended by increasing the allowed fap or by using a targeted search . by comparison
, the semi - coherent matched filtering algorithm @xcite can reach @xmath102 gpc for an overall fap of @xmath77 , but at a presently undetermined detection rate ( perhaps @xmath100 ) . broadly speaking
, this time frequency search has better than half the reach of the semi - coherent search @xcite , but at a tiny fraction of the computational cost .
this is very encouraging , given the simplicity of the technique , and suggests that a method like this could be a valuable first step for detecting the loudest emri events in the lisa data .
the approach is very similar to the ` excess power ' technique used in ligo @xcite , which was designed to search for bursting signals .
although we have concentrated on emris , this technique will also be sensitive to other types of source , including bursts , galactic white dwarf binaries etc .
and so it may have many applications .
however , one significant issue that we have not yet considered is that of confusion .
the lisa data will be dominated by a superposition of signals from many different astrophysical sources .
so far , we have only examined the problem of detecting a single source in instrumental noise ( plus an astrophysical foreground of white dwarf binaries ) .
this algorithm should be able to make detections when there are multiple sources if the sources are reasonably well separated in time and frequency or of widely different signal to noise ratios , but it is not clear how it will fair on overlapping sources of comparable brightness .
it is likely that the performance will be significantly impaired in such cases , due to the simplicity of the algorithm .
additionally , a shortcoming of this simple technique is that while it can tell us that a source is present , it does not provide much information about the properties or parameters of that source . in principle
, a time - frequency analysis can provide at least the frequency and rate of frequency drift of an inspiral source , but this requires more sophisticated algorithms .
one possibility would be to use boxes of different ( i.e. , non - rectangular ) shapes , or to use a pattern recognition algorithm in the ( binned or unbinned ) t - f plane to search for tracks corresponding to inspirals .
these more sophisticated techniques should also cope more readily with confusion .
finally , it will be instructive to test this and future algorithms using more accurate inspiral waveforms .
the simple kludged waveforms used here are purely quadrupolar , and so their frequency structure is not as complex as true emri waveforms which might influence our results . despite these remaining issues , the performance of this algorithm is very promising and suggests that a time - frequency method of some kind is likely to be a useful first step in detecting emris and other types of source in the lisa data .
this provides a good incentive for developing and testing t - f algorithms in the ways suggested above and this work is now underway .
we thank curt cutler , teviet creighton , leor barack and kip thorne for critical discussions of this work .
this work was supported in part by the max planck institut fuer gravitationshphysik ( albert einstein institut ) ( lw ) , by nasa grants nag5 - 12834 and nag5 - 10707 ( jg ) and by st .
catharine s college , cambridge ( jg ) .
10 wen l and gair j r 2005 s445 gair j r , barack l , creighton t , cutler c , larson s l , phinney e s and vallisneri m 2004 s1595 barack l and cutler c 2004 d * 69 * 082005 nelemans g , yungelson l r and portegies zwart s f 2001 _ astronomy & astrophysics _ * 375 * 890 timpano s e , rubbo l j and cornish n j 2005 _ preprint _ gr - qc/0504071 cutler c 1998 d * 57 * 7089 glampedakis k , hughes s a and kennefick d 2002 d * 66 * 064005 creighton t , gair j r , hughes s a and vallisneri m 2005 , in preparation gair j r and glampedakis k 2005 , in preparation cornish n j and rubbo l j 2003 d * 67 * 022001 , erratum - ibid .
d * 67 * 029905 rubbo l j , cornish n j and poujade o 2004 d * 69 * 082003 vallisneri m 2005 d * 71 * 022001 madau p and rees m j 2001 _ astrophys . j. lett .
_ * 551 * l27 miller m c 2002 _ astrophys . j. _ * 581 * 438 anderson w g , brady p r , creighton j d e and flanagan e e 2001 d * 63 * 042003 | the inspirals of stellar - mass compact objects into supermassive black holes constitute some of the most important sources for lisa .
detection of these sources using fully coherent matched filtering is computationally intractable , so alternative approaches are required . in a previous paper @xcite
, we outlined a detection method based on looking for excess power in a time - frequency spectrogram of the lisa data .
the performance of the algorithm was assessed using a single ` typical ' trial waveform and approximations to the noise statistics . in this paper we present results of monte carlo simulations of the search noise statistics and examine its performance in detecting a wider range of trial waveforms .
we show that typical extreme mass ratio inspirals ( emris ) can be detected at distances of up to 13 gpc , depending on the source parameters .
we also discuss some remaining issues with the technique and possible ways in which the algorithm can be improved . |
it is widely believed that feedback from active galactic nuclei ( agn ) during the epoch of galaxy formation is required to explain the relation between black hole and bulge mass / velocity dispersion @xcite , the deficit of bright galaxies in the galaxy luminosity function @xcite , and the completion of star formation in massive galaxies at epochs of redshift @xmath2 @xcite .
it is envisaged that either radiation or outflows from a galactic nucleus impedes the infall of star - forming gas once the central black hole grows to a critical size .
accordingly , @xcite , @xcite , @xcite , and others , have appealed to the physics of either energy - driven or momentum - driven bubbles in order to explain the relationships between the mass of the black hole and the parameters of the host galaxy . in order to model the galaxy luminosity function ,
@xcite have utilized semi - analytic models based on the output of the millennium simulation , incorporating radio - mode feedback , coupled with a prescription for the accretion rate into the center of each evolving galaxy .
their feedback prescription is motivated by the well - documented evidence for the effect of radio galaxies on cooling flow galaxies @xcite .
there is also a growing literature on cosmological simulations in a @xmath3cdm cosmogony involving both dark matter and gas dynamics , which incorporate feedback from both supernovae ( sn ) and black holes , and which test scenarios of galaxy merging and growth @xcite . in simulations using the gadget-2 code
@xcite the total number of sph particles representing the baryonic component exceeds 250 million , and the effective spatial dynamic range is an equally impressive @xmath4 per dimension .
nevertheless , the best spatial resolution is about 2 kpc and does not resolve the spatial scales where important dynamical processes occur .
this is highlighted by the prescriptions for black hole growth and feedback described , for example in the work by @xcite .
accretion is described in terms of the bondi - hoyle accretion rate multiplied by a factor , which can be as large as @xmath5 .
the rationale for this approach is that at sub - grid scales the density would be larger , and the real accretion rate would be appreciably higher . however , the higher densities and the consequent cooling and fragmentation on sub - grid kpc scales would create a multi - phase interstellar medium ( ism ) , the physics of which is not satisfactorily captured by the simulations . in particular , such a medium is porous , and the interaction between jets and the dense , potentially star - forming clouds of gas is complex , with radio - emitting plasma being able to channel through holes in the density distribution , rather than isotropically impacting a smooth distribution of dense gas as shown in previous work @xcite . given that black - hole driven
feedback occurs in bright galaxies , there does not appear to be a consensus on the type of feedback : how much agn power is involved , and does the feedback involve radiative or mechanical processes or both ? in their radio - mode model , @xcite
attribute feedback to radio galaxies accreting at rates well below eddington , and for typical ellipticals , this means low - powered fanaroff - riley class i radio galaxies are the primary drivers of feedback . in @xcite the momentum for dispersing the circumnuclear gas comes from a quasar wind ; in @xcite the momentum of the outflow is provided by an eddington - limited radiation - driven wind ; in sph simulations the accretion rate , which is the ultimate power source for an outflow , can approach eddington values @xcite . there is also an issue of what class of agn actually drives black hole feedback .
the models by @xcite invoke low - powered radio galaxies .
observers , however , have focused on powerful radio galaxies , mainly at @xmath6 ; in these galaxies there is evidence for substantial outflows of line - emitting gas and neutral gas driven by the radio jets @xcite .
it is possible that there is a role for radio galaxies with a range of powers : powerful radio galaxies may be responsible for the establishment of the stellar mass at @xmath6 , and less powerful sources may be responsible for the maintenance of the stellar content through the inhibition of cooling flows @xcite .
we also note that radio galaxies are mainly relevant to the elliptical galaxy population , and that the separate luminosity functions for early and late - type galaxies @xcite indicate the requirement for feedback in both populations .
in this paper we consider the potential role of radio galaxies in agn feedback and address the following questions : ( 1 ) what jet power is required for the radio galaxy phase to have an important effect on inhibiting star formation in a given host , and ( 2 ) is the range of radio powers broad enough that radio galaxies could affect the entire distribution of bright ellipticals ? in this paper we present progress in answering these questions through simulations with a resolution of @xmath7 per computational cell .
these simulations confront the sub - grid physics that current large scale sph simulations do not address . in particular , we consider the effect of powerful relativistic jets on a two - phase ism consisting of hot gas , in which is embedded a dense porous phase of warm gas .
these simulations extend the simulations described in @xcite , in which a jet with a kinetic power of @xmath8 propagates through an inhomogeneous medium in the form of an almost keplerian fractal disk .
it is evident from that simulation that in the geometry considered , jets of that power could not exert enough impact on the clouds to disperse them , and that the jets would not have a important effect on star formation in the core of the host galaxy . in the present simulations we consider jets with powers ranging from @xmath0 to @xmath1 propagating through a two - phase medium , in which the dense clouds are spherically distributed throughout a region @xmath9 in diameter .
these initial data are meant to describe a typical protogalaxy , in which dense gas has accumulated in the core .
the fractal distribution of the dense gas enables us to directly examine the effect of porosity on the evolution of potentially star - forming clouds . in the following sections we describe the parameters of the simulations in more detail and then discuss our results
in our simulations , we use the publicly available , open - source code flash @xcite version 3.2 , to which we have added code to incorporate radiative cooling of thermal gas and code to advance advected scalars in the relativistic hydrodynamic solver .
details of the solver are in @xcite .
we exploit the adaptive mesh capabilities of flash , utilizing up to seven levels of refinement ; this corresponds to a nominal cubical simulation grid of @xmath10 in physical dimensions , consisting of @xmath11 cells at maximum resolution .
however , note that a restricted one parameter scaling of physical dimensions is possible @xcite .
the relevant jet parameters , which are initiated and maintained constant at the boundary that is the jet inlet , are : the lorentz factor @xmath12 , where the @xmath13 is the jet velocity in units of the speed of light ; the proper density parameter , @xmath14 , where @xmath15 is the polytropic index , @xmath16 is the rest mass density , and @xmath17 is the pressure ; and the jet power . let @xmath18 be the jet cross - sectional area .
the jet power is @xmath19 we adopt @xmath20 for both jet and ambient gas . in all simulations presented here , @xmath21 , and @xmath22 .
the jet inlet is a circular region of area @xmath18 centered at @xmath23 with normal @xmath24 .
the initial jet velocity is parallel to the @xmath25-axis .
the boundary @xmath26 is reflective , apart from the jet inlet .
all other boundaries of the simulation domain are designated as inflow / outflow boundaries .
the two parameters describing the hot phase of the ism are the temperature , @xmath27 , which is fixed at @xmath28 , and the total number density @xmath29 .
the pressure , @xmath30 , where @xmath31 is boltzmann s constant , for the ism in each of the simulations is either @xmath32 or @xmath33 , typical of the ism in giant elliptical galaxies @xcite .
the distribution of dense , warm clouds is prescribed in similar fashion to that in @xcite , following work on terrestrial clouds by @xcite .
the main difference is that , here , the average mass density is uniform , whereas in @xcite the average density is that of a near - keplerian disk . for completeness , we describe the other details of the cloud distribution used in this series of simulations : 1 .
we begin by constructing a cube of density fluctuations in which the single point statistics of the density distribution are described by a log - normal distribution .
the mean , @xmath34 , of the parent distribution is 1.0 , and the variance , @xmath35 , these values being consistent with starburst reddening and extinction models @xcite .
this value of @xmath36 is the same as that used in @xcite . in numerical simulations of supersonic turbulence ,
@xcite find @xmath37 and 35 for solenoidal ( divergence free ) and compressive ( curl free ) forcing respectively , so that our adopted value of 5 is closer to their solenoidal result .
the power spectrum of the density distribution @xmath38 for the range of waves numbers @xmath39 and zero outside of this range . the parameter @xmath40 sets the maximum cloud size . for these simulations @xmath41 ( in cell units in fourier space ) and this value limits the maximum size of an individual cloud to approximately @xmath42 for 512 cells along each axis of the @xmath9 computational cube . limiting the cloud size to @xmath42 allows for appropriate coverage and variation of hot phase ism and warm phase ism along any path through the region of our simulation grid filled with clouds .
this ensures that the spatial distribution of clouds is approximately isotropic , and that the jet plasma encounters a statistically significant numbers of clouds along its main axis of propagation . as a result ,
the flow of the jet plasma through the hot phase is approximately isotropic .
the parameter @xmath43 is set by the resolution of the simulation .
here @xmath44 and is equivalent to two computational cells . 3 .
the parent , unit mean distribution is scaled by the mean particle density of warm clouds , @xmath45 ; this parameter is determined by the ratio of @xmath45 to @xmath29 .
the temperature in each cell of the warm gas distribution is determined by pressure equilibrium with the hot gas .
when the warm gas temperature exceeds @xmath46 ( in the lowest density parts of the distribution ) , it is deemed to be thermally unstable and is replaced by hot gas .
this makes the gas porous .
we can then define a volume filling factor , @xmath47 , of the warm phase by integrating over the log - normal density distribution above the critical density @xmath48 , where @xmath49 is the mean molecular weight and @xmath50 is an atomic mass unit ( see also appendix b of * ? ? ?
+ for a given value of @xmath51 , the volume filling factor is determined by the ratio of the mean warm phase density to hot phase density , @xmath52 ; the larger this parameter the larger the filling factor .
this log - normal , fractal , porous distribution of clouds is an important feature of the simulations presented here .
+ @xcite and @xcite identified porosity as a key parameter in their models of feedback - regulated star - formation , since the porosity determines the extent to which energy - driven bubbles generated by sn or agn activity are confined .
the porosity in the two - phase ism employed in our simulations plays a similar role , in that it constrains the progress of the jet along its principal axis and ensures the confinement of the pressurized bubble for times much larger than the dynamical time of an unimpeded jet .
+ the porosity of dense clouds is the primary reason for the differences in the evolution of the radio source and the differences in energy and momentum imparted by the jet to the ism between our simulations and those of a jet propagating into a uniform medium . 5 .
the radial extent of the gas is truncated by a sphere of radius @xmath53 .
hence , we are simulating the interaction of jets with warm gas within 12 core radii of the parent galaxy .
this justifies the neglect of gravity .
we include non - equilibrium , optically thin atomic cooling for @xmath54 @xcite ; for @xmath55 the cooling is set to zero .
we use updated solar abundances @xcite .
+ the non - equilibrium , cooling function is pre - calculated for a high temperature ( @xmath56 ) shock . the main reason for using such a function was outlined in 2.1 of @xcite and , for completeness , we summarize it here : following a shock there is a short - lived cooling spike as low - ionization material is suddenly shocked to a high temperature . in a numerically well - resolved shock
this phase is short - lived and provides about 1% of the total cooling .
however , in a simulation in which each shock is not well - resolved ( as is the case here ) this spike can dominate the cooling and the situation is exacerbated by the interpolation of intermediate temperatures at each cell . apart from the initial ( and unimportant ) cooling spike ,
our adopted non - equilibrium cooling function is a good approximation to the cooling function of shocks for velocities above @xmath57 .
since the cooling function is pre - calculated and interpolated within the code there are no additional computational costs associated with a non - equilibrium calculation .
the utilization of an _ optically thin _
cooling function also requires some comment , given the dense , reasonably extended regions that evolve within the simulations . consider a region with electron density @xmath58 and thickness @xmath59 .
the electron scattering optical depth @xmath60 . in the simulations with the densest gas the initial average electron density @xmath61 so that the largest clouds with @xmath62 would be optically thin but verging on optically thick where the density is a factor of 10 above average . as a result of the radiative shocks , the electron density increases by a factor of @xmath63 .
hence , even some of the less dense regions could become optically thick to scattering .
scattering alone is not enough to invalidate the assumption of optically thin cooling , and an admixture of dust is required to cause absorption . moreover , the cooling in such regions occurs before maximum density has been obtained , and photons emitted by cooling plasma following a radiative shock have an escape route through the lower density regions of the shock .
in addition , the temperature of the gas in the cooling , optically thin region of a radiative shock would be so high that dust would be destroyed .
hence , optically thin cooling is a reasonable first approximation , which _ may _ be limited in the largest clouds .
more refined radiation hydrodynamic simulations , possibly involving larger clouds , may need to take optical depth effects into account .
ccccccc a & 45 & 0.1 & @xmath32 & & & + b & 46 & 1.0 & @xmath33 & 1000 & 0.42 & 16 + c & 46 & 0.1 & @xmath32 & 100 & 0.42 & 1.6 + c@xmath64 & 46 & 0.1 & @xmath32 & 30 & 0.13 & 0.32 + d & 45 & 1.0 & @xmath33 & 1000 & 0.42 & 16 + d@xmath64 & 45 & 1.0 & @xmath33 & 300 & 0.13 & 3.2 + e & 45 & 0.1 & @xmath32 & 100 & 0.42 & 1.6 + e@xmath64 & 45 & 0.1 & @xmath32 & 30 & 0.13 & 0.32 + f & 44 & 0.1 & @xmath32 & 100 & 0.42 & 1.6 + f@xmath64 & 44 & 0.1 & @xmath32 & 30 & 0.13 & 0.32 + g & 44 & 1.0 & @xmath33 & 1000 & 0.42 & 16 + g@xmath64 & 44 & 1.0 & @xmath33 & 300 & 0.13 & 3.2 + h & 43 & 0.1 & @xmath32 & 100 & 0.42 & 1.6 in summary , the key model parameters used in the different simulations are the jet power and the densities of the hot and warm ism phases , which determine the cloud filling factor .
table [ t : parameters ] summarizes the assigned and derived parameters that we used for our simulations .
the panels in figs . [ f : densitya ] and [ f : densitye ] show the evolution of the density in the two simulations a and e , respectively . the figures present slices through the mid - plane @xmath65 of the grid . a is a reference simulation of a jet and homogeneous ism .
the progress of such a jet has been well established over two decades of research , although we do note the unstable jittering of the three - dimensional jet once it has traversed about 0.5 kpc ( @xcite , also evident in @xcite ) , which leads to multiple hot spots . the progress of the jet in fig .
[ f : densitye ] ( simulation e ) is completely different : initially the jet is deflected in various directions as it floods through the porous screen of dense clouds , finding channels of least resistance and gradually dispersing the clouds through the effect of the ram pressure of the non - thermal plasma . in addition , the slower rate of progress of the jet traps the high pressure cocoon material ; a pseudo - spherical bubble is driven into the ism , and the dense clouds are driven outwards and dispersed as this bubble expands . some of the dispersed material is accelerated to speeds @xmath66 while the speed of the densest parts of the clouds reaches several hundred @xmath67 .
the lateral extent of the cocoon of the a and e jets is very different .
when the cloud - free jet has progressed about 0.4 kpc its radial extent is about 0.2 kpc , whereas the radial extent of the cocoon from the cloud - impeded jet is about 0.4 kpc .
for the same length the latter jet processed a factor of 8 larger volume of the ism , removing one of the reservations about the relevance of powerful radio galaxies to agn feedback .
note also the streams of low density , pressurized cocoon material shooting out in random directions .
this is a direct result of the porous ism , into which the main jet is propagating .
figures 2.12.5 and corresponding mpeg animations depicting the density evolution of runs e , c@xmath64 , d@xmath64 , g , and h , respectively , are available in the electronic edition of the _
astrophysical journal_. we have generated images of synthetic radio surface brightness , based on the density of the non - thermal plasma and also assuming that the magnetic field is proportional to the pressure and a spectral index of 0.6 .
a set of snapshots from simulation e are shown in fig .
[ f : surfbrite ] .
the interaction between jet and clouds produce a markedly different morphology from that of a classical radio galaxy .
images such as these can be informatively compared with images of young gps and css radio galaxies ( cf .
a convenient approach for assessing the effects of agn feedback on galaxy formation is to examine the velocity imparted to gas , which could form new stars .
the usual criterion for inhibition of further galaxy formation is that the velocities imparted to the clouds is greater than the velocity dispersion , @xmath68 , of the host galaxy @xcite .
this does not necessarily mean that the clouds would be ejected to large radii but it does mean that the clouds would be highly dispersed within the potential well of the host galaxy . in order to track the different gas components we use tracers , which are the mass concentration of that component in each cell .
in particular we use a warm gas tracer , @xmath69 , which is initialized to unity in each cell of warm thermal gas .
figure [ f : v_clouds ] shows the velocity of the dispersed warm interstellar material , for @xmath70 , in simulation e at various epochs . at @xmath71
the maximum mean cloud velocity is reached .
there is a wide range in velocity with the remaining dense cloud cores showing the lowest velocities , material resulting from the dispersal of the clouds showing intermediate velocities @xmath72 , and lighter material showing even higher velocities @xmath73 .
figures 4.14.5 and corresponding mpeg animations that depict the velocity evolution of the warm material , for which @xmath70 , in runs e , c@xmath64 , d@xmath64 , g , and h , respectively , are available in the electronic edition of the _
astrophysical journal_. taking @xmath74 , the density in a cell , and @xmath75 , the radial velocity in a cell , we compute for each simulation the mass - weighted mean radial velocity over all @xmath76 cells defined by @xmath77 we adopt @xmath78 as the criterion for inhibition of further galaxy formation .
this raises the question : what value of @xmath68 is appropriate ? if we use a value relevant to a giant elliptical @xmath79 , this discriminates against lower mass galaxies ( @xmath80 say ) , in which jet inhibition of star formation may be important in earlier epochs of galaxy formation . in order to establish the relevance of these simulations for all phases of galaxy formation we adopt the following procedure , in which the jet power is parametrized by its ratio @xmath81 relative to the eddington luminosity , that is , @xmath82 , where @xmath83 is the proton mass and @xmath84 is the thomson scattering cross section . adopting the magorrian relation @xcite between black hole mass and velocity dispersion , @xmath85 , where @xmath86 is the velocity dispersion in units of @xmath87 , gives the jet power @xmath88 , and the velocity dispersion expressed in terms of the jet power in units of @xmath89 is @xmath90 we plot the results from each simulation on a diagram , fig . [
f : v - p ] , of the maximum value of @xmath91 vs @xmath92 . also plotted are the loci of @xmath68 vs @xmath92 for incremental values of @xmath93 , using equation ( [ e : sigma - p ] ) . for
a given value of @xmath81 a jet will disperse the warm cloudy material if the corresponding @xmath94 point lies above the specific @xmath95 locus .
for example , consider the point for simulation f represented by the point @xmath96 , @xmath97 . if @xmath98 , the cloud speed comfortably exceeds the corresponding velocity dispersion of approximately @xmath99 .
on the other hand , if @xmath100 then the inferred velocity dispersion is approximately @xmath101 , and it is marginal whether feedback from such a jet would influence the evolution of the host galaxy . for @xmath102 , we expect no substantial feedback .
a number of patterns are immediately obvious from this diagram .
consider first the sequence of simulations cefh .
these have the same values of @xmath103 and warm cloud density , @xmath45 , and are ordered by decreasing jet power from @xmath1 down to @xmath104 .
all of these jets can disperse cloudy material for @xmath105 but for @xmath106 the jets would not strongly influence subsequent star formation .
next consider the bdg sequence .
these simulations have a higher @xmath103 and the same ratio of cloud to ism densities and , thus , the same filling factor .
the clouds are denser and harder to move , and in this sequence the critical value of the jet power to eddington ratio , @xmath81 , is higher at around @xmath107 .
we have also conducted simulations , in which the filling factor is reduced by reducing the average density of the clouds for a fixed @xmath103 .
these are the sequences c@xmath64e@xmath64f@xmath64 and b@xmath64d@xmath64g@xmath64 plotted in fig .
[ f : v - p ] . at high jet powers ,
these lower filling factor simulations all exhibit higher values of @xmath108 , that is , it is more difficult for the jet to disperse the clouds when the clouds are more porous .
this is the result of the non - thermal plasma being able to escape more readily through a more porous medium . on the other hand , the simulations f@xmath64 and g@xmath64 at jet powers of @xmath109 occupy similar positions on the @xmath91@xmath92 diagram as their higher filling factor counterparts . in these cases
the lower power jets do not break through as readily , and the confinement time of the non - thermal gas is sufficient to build up a large pressure , which accelerates the clouds .
the results for even lower filling factors at all jet powers will be of interest but requires higher resolution simulations to prevent the clouds from becoming too pixelated .
another aspect of jet feedback , which is readily interpreted using this diagram is the relative importance for feedback of jets of different power .
compare the points h and c in fig .
[ f : v - p ] .
these points have approximately the same @xmath108 . however , the _ maximum _ value of the velocity dispersion , for which the jet in simulation h would satisfy the feedback criterion , @xmath78 , is about @xmath110 whereas the corresponding value for c is about @xmath111 .
thus , we can make the qualitative remark that low - powered ( @xmath112 ) jets are relevant to feedback in less massive hosts , that is , galaxies in the early stages of hierarchical merging .
do stars form as a result of the excess pressure in the radio lobes ?
let @xmath113 be the ambient pressure surrounding a sphere with central number density @xmath114 and temperature @xmath115 . if magnetic fields are neglected , considerations of star formation focus on the bonnor - ebert mass @xcite @xmath116 or the 50% larger critical
mass @xmath117 derived from the virial theorem for a uniform density gravitating mass @xcite . for a mass in excess of @xmath118 ,
say , we expect that pressure driven gravitational instability will lead to star formation @xcite . in the pre - jet phase of our highest pressure simulations ( @xmath119 )
the critical mass is approximately @xmath120 compared to the mass , @xmath121 , of the largest clouds with radius , @xmath122 .
hence , at this stage , the mass of the largest clouds is less than the critical mass , although not by a great amount .
what is the effect on star formation when the ambient pressure increases as a result of the formation of the high - pressured radio lobe ?
for example , in simulation f@xmath64 , the value of @xmath30 , within the lobe rises from @xmath123 to about @xmath124 , the critical mass decreases to about @xmath125 and jet induced starformation may be feasible .
however , the overpressured radio lobe is a highly turbulent environment and it is also helpful to compare the collapse timescale of the cloud , @xmath126 to the ablation time scale @xmath127 , where @xmath128 is the ablation velocity of cloud material .
clouds are destroyed on a time scales of a few @xmath129 . if we allow the cloud density to increase by a maximal factor of 100 as a result of radiative shocks and adopt fiducial values for the other parameters we have @xmath130 and @xmath131 . for
this extreme case one may expect star formation to occur . on the other hand ,
a factor of 100 increase in density is high so that we expect that most clouds would be dispersed before they could gravitationally collapse to form stars .
a similar conclusion was reached by @xcite in their study of the effect of radio - lobes driven by jets with powers ranging from @xmath132 to @xmath1 interacting with a single cloud with a radius of @xmath133 .
given these estimates , under what circumstances would we expect _ jet - induced _ rather that _ jet - inhibited _ star formation ? first
, if the temperature of the clouds were lower than @xmath134 the critical mass for gravitational instability ( @xmath135 ) would also be lower .
second , we can also speculate that the clouds towards the edge of the radio lobe would not be so strongly affected by the lower level of turbulence in that region but they are still initially affected by the overpressure of the radio lobe bow - shock .
therefore it is possible that there may be some star formation in these regions .
these issues are beyond the scope of this paper but they should be of interest in future work . panels a and b in fig .
[ f : combi ] show the kinetic energy imparted to the clouds in each simulation and the efficiency of the transfer of energy from the jet to the clouds , respectively . in each simulation
the efficiency is quite high and , except for runs c@xmath64 and e@xmath64 , the efficiency is still increasing by the end of the simulation .
all curves show a slight dip or inflection before steeply rising ; this feature occurs at the point where the jet plasma breaks out of the confining clouds .
( in the simulations of lower powered jets , the break out is smoothed out because the non - thermal plasma breaks out everywhere almost simultaneously . )
these curves show a decrease in efficiency with decreasing value of the ratio of jet power to cloud density ( @xmath136 ) . from the curves of @xmath91 as a function of time shown in panel c of fig .
[ f : combi ] , we see that the maxima in @xmath91 occur just before jet break - out , which releases some of the pressure in the quasi - spherical bubble .
compare the similarity of the evolution of @xmath137 in b and e , d and f , g and h , b@xmath64 and e@xmath64 and d@xmath64 and f@xmath64 .
these pairs of runs have the same values of @xmath136 and exhibit comparable maximum radial velocity dispersion .
the rate of momentum transfer from the jet to the clouds can be an issue in this field , to which @xcite drew attention .
for example , the range of cloud and jet parameters presented in @xcite indicates that the cloud momentum flux may exceed that of the jet .
hence , it is useful to examine the momentum budget to see what can be expected from the observations .
momentum is not as straightforward as energy since the momentum in a volume surrounding a moving cloud is affected by the pressure ( and magnetic stresses ) integrated over the bounding surface of that volume . in this case the shocked jet gas provides a large pressure as a result of the _ energy _ flux in the jet .
therefore , in panel d of fig .
[ f : combi ] , we present the mechanical advantage ( the ratio of the momentum in clouds to the jet momentum ) as a function of time in each simulation . in each case
the mechanical advantage exceeds unity .
note that the greatest mechanical advantage is achieved by the lowest power jets .
low powered jets `` punch through '' the warm clouds more slowly , and the shocked jet gas spreads more effectively through the volume of warm clouds and has a greater relative effect on the cloud momentum . in each of the simulations , the jet plasma , after a sufficiently long time , escapes the cloud region . the more powerful the jet is , and the lighter the clouds are , the more centrally collimated the outbreak of the plasma is .
we end the simulations when the jet plasma reaches the domain boundaries , and boundary effects become noticeable .
we may , nonetheless , speculate on the subsequent evolution of the system . in the highest power jet simulations ,
we see the jet breaking through along its principal propagation axis ( @xmath26 ) and piercing the energy bubble that was inflated during the flood - and - channel phase of evolution . except for runs g and h
, the jets in all our simulations will eventually emerge in a similar fashion , inflating a second , more classically shaped radio lobe beyond the first quasi - spherical bubble to a distance depending on the jet power . in runs g and h
, the jet may be trapped indefinitely in the cloud region .
while the common notion is that hierarchical radio lobe structures are produced by recurrent jet activity @xcite , we note that they may also be produced by the randomized emergence of radio plasma from an inhomogeneous , porous ism .
after the jet escapes along its principal propagation axis , the efficiencies in transfer of momentum and energy to the clouds drop quickly . this stage
is already seen in runs c@xmath64 and e@xmath64 , and will occur at later times in the other runs ( except for g and h , if the jets in those runs remains trapped for the duration of their activity ) . while the clouds in our simulations are dispersed to high velocities , the host galaxy will retain most of the cloud material , except for a diffuse component comprising the high velocity tail in the distribution , which is concentrated along the jet axis near the region where the central jet breaks out . by the end of the runs , the clouds in all simulations have reached and passed their maximum value of @xmath91 .
the suite of simulations presented in this paper strongly reinforce the importance of inhomogeneity in the consideration of jet interactions with the ism @xcite .
inhomogeneity has several important consequences : ( 1 ) it affects the early morphology of the radio source as a result of the interaction of the jet and lobe with the obstructing clouds .
( 2 ) the radio source affects a much larger volume of the host galaxy because of the channeling of the jet flow in different directions .
( 3 ) the shocked jet clouds are left in the wake of the non - thermal plasma and , as first noted by @xcite , would continue to emit shock excited line emission with the shocks driven by the high pressure gas in the non - thermal cocoon ; this shock - excited emission is in addition to the emission that may be driven by photoionization by the nucleus .
( 4 ) the porosity of dense gas determines the ease with which a jet in a given host can disperse this gas , which is a potential source of new stars . in powerful sources , higher porosity gas is less easily dispersed but this trend is not evident for lower - powered ( @xmath138 ) jets .
( 5 ) jets of all powers can exert a considerable feedback effect on their host galaxies , although lower - powered jets only play a role in the lower velocity dispersion hosts .
brighter galaxies require more powerful jets to disperse dense clouds .
( 6 ) the efficiency of transfer of kinetic energy from the jet to the dense gas is high .
( 7 ) the efficiency of transfer of momentum to the clouds is also high with mechanical advantages considerably exceeding unity .
inhomogeneity is therefore crucial when considering agn feedback on the kiloparsec scale , both for the interpretation of radio and optical emission - line morphology in radio sources , which may be generating feedback , and for incorporating the effect of jet - mediated feedback on host galaxies of different size into large scale simulations , in which the resolution @xmath139 .
an important conclusion from these simulations is that jets with eddington efficiency @xmath106 are unlikely to have an effect on evolving galaxies when the pre - starforming gas exists in the form of clouds , which are relatively dense and cool compared to the hot ism .
this critical value of @xmath81 is relevant for clouds with a high filling factor of 0.42 and a value of @xmath140 for the ism .
we have shown that when the filling factor decreases or the pressure of the ism increases the critical value of @xmath81 increases , and values of @xmath141 are not unrealistic .
the precise values of @xmath108 will have to await further higher resolution simulations with high porosity dense gas .
this research was undertaken on the nci national facility at the australian national university .
the software used in this work was in part developed by the doe - supported asc / alliance center for astrophysical thermonuclear flashes at the university of chicago .
we are grateful to prof .
mitchell begelman for detailed comments on an initial version of the manuscript and to dr .
ralph sutherland who provided revised cooling functions , incorporating the latest solar abundances , for this work .
we are grateful to our anonymous referee for a thorough reading of the manuscript and for a number of constructive comments , which have improved the presentation . , v. & silk , j. 2008 , , 389 , 17501762 , m. , grevesse , n. , & sauval , a. j. 2005 , september ) , volume 336 of astronomical society of the pacific conference series 25+ , r. & saglia , r. p. 1999
, august ) , in galaxy dynamics - a rutgers symposium , ed .
d. r. merritt , m. valluri , & j. a. sellwood , volume 182 of astronomical society of the pacific conference series 113+ , w. b. 1956 , , 116 , 351359 , c. m. & schaye , j. 2009a , , 398 , 5374 , c. m. & schaye , j.arxiv e - prints , , s. , et al .
2001 , , 326 , 255273 , d. j. , et al .
2006 , mnras , 365 , 1128 , m. a. , fischera , j. , groves , b. , sutherland , r. s. , kewley , l. j. , tuffs , r. , popescu , c. , & leitherer , c. 2004 , in the interplay among black holes , stars and ism in galactic nuclei , ed
. t. storchi - bergmann , l. c. ho , & h. r. schmitt , volume 222 of iau symposium 313314 , r. 1955 , zeitschrift fur astrophysik , 37 , 217232 , a. c. 1999 , , 308 , l39l43 , a. c. , sanders , j. s. , allen , s. w. , crawford , c. s. , iwasawa , k. , johnstone , r. m. , schmidt , r. w. , & taylor , g. b. 2003 , september ) , , 344 , l43l47 , c. , roman - duval , j. , klessen , r. s. , schmidt , w. , & mac low , m .- m .
2010 , march ) , , 512 , a81 + , j. , dopita , m. a. , & sutherland , r. s. 2003 , apjl , 599 , l21l24 , b. , et al .
2000 , november ) , , 131 , 273334 , k. , et al .
2000 , apjl , 539 , l13 , j. , tadhunter , c. , morganti , r. , bellamy , m. , gonzlez delgado , r. m. , tzioumis , a. , & inskip , k. j. 2006 , , 370 , 16331650 , j .- s . , glazebrook , k. , cowie , l. l. , & tinney , c. 2003 , , 584 , 203209 , a. 2005 , , 635 , l121l123 , m. & gaibler , v. 2010 , in agn feedback in galaxy formation , ed . v. antonuccio - delogu & j. silk in press , m. r. & mckee , c. f. 2005 , , 630 , 250268 , g. m. & austin , p. h. 2002 , in 11th conference on atmospheric radiation , ed .
g. h. smith & j. p. brodie , american meteorological society conference series 123126 , p. , ferguson , h. c. , dickinson , m. e. , giavalisco , m. , steidel , c. c. , & fruchter , a. 1996 , , 283 , 13881404 , j. , et al .
1998 , aj , 115 , 22852305 , w. g. & brighenti , f. 2003 , , 41 , 191239 , w. h. 1957 , , 117 , 562 , b. r. , nulsen , p. e. j. , wise , m. w. , rafferty , d. a. , carilli , c. , sarazin , c. l. , & blanton , e. l. 2005 , january ) , , 433 , 4547 , a. & bodo , g. 2005 , november ) , , 364 , 126136 , a. , plewa , t. , & bodo , g. 2005 , september ) , , 160 , 199219 , y. , hardee , p. , & nishikawa , k .-
2007 , june ) , , 662 , 835850 , r. , holt , j. , tadhunter , c. , & oosterloo , t.arxiv e - prints , , n. p. h. , et al .
2009 , , 395 , l16l20 , p. , et al .
2002 , , 336 , 907931 , p. , jones , c. , forman , w. , churazov , e. , mcnamara , b. , david , l. , & murray , s.arxiv e - prints , , d. j. & jamrozy , m. 2009 , september ) , bulletin of the astronomical society of india , 37 , 6389 , c. j. , bicknell , g. v. , sutherland , r. s. , & midgley , s. 2005 , mnras , 359 , 781800 , j. , et al .
2010 , , 402 , 15361560 , p. a. , wall , j. v. , kellermann , k. i. , jackson , c. a. , & hawkins , m. r. s. 1996 , , 384 , 439441 , j. 1997 , may ) , , 481 , 703+ , j. 2001 , june ) , , 324 , 313318 , j. & norman , c. 2009 , july ) , , 700 , 262275 , j. & rees , m. j. 1998 , a&a , 331 , l1l4 , v. 2005 , mnras , 364 , 11051134 , v. & hernquist , l. 2003a , , 339 , 289311 , v. & hernquist , l. 2003b , , 339 , 312334 , r. s. & bicknell , g. v. 2007 , apjs , 173 , 3769 , r. s. , bicknell , g. v. , & dopita , m. a. 2003 , apj , 591 , 238257 sutherland , r. s. & dopita , m. a. 1993 , apjs , 88 , 253 , s. , et al .
2002 , apj , 574 , 740753 | over cosmic time , galaxies grow through the hierarchical merging of smaller galaxies . however , the bright region of the galaxy luminosity function is incompatible with the simplest version of hierarchical merging , and it is believed that feedback from the central black hole in the host galaxies reduces the number of bright galaxies and regulates the co - evolution of black hole and host galaxy .
numerous simulations of galaxy evolution have attempted to include the physical effects of such feedback with a resolution usually exceeding a kiloparsec .
however , interactions between jets and the interstellar medium involve processes occurring on less than kiloparsec scales . in order to further the understanding of processes occurring on such scales , we present a suite of simulations of relativistic jets interacting with a fractal two - phase interstellar medium with a resolution of two parsecs and a largest scale of one kiloparsec .
the transfer of energy and momentum to the interstellar medium is considerable , and we find that jets with powers in the range of @xmath0@xmath1 can inhibit star formation through the dispersal of dense gas in the galaxy core .
we determine the effectiveness of this process as a function of the ratio of the jet power to the eddington luminosity of the black hole , the pressure of the interstellar medium and the porosity of the dense gas . |
attempts to understand stress distribution in static , non - cohesive , granular materials have uncovered a rich structure of force chains that currently is poorly understood @xcite .
although mechanical systems such as bead or granular packings subject to uniaxial compression seem straightforward , they harbor fundamental theoretical challenges concerning the connection between the micro and macro scales . such systems are far from thermal equilibrium since thermal energy scales are negligible compared to the potential energies stored in elastic deformation of the grains . in addition , the force between grains is strongly nonlinear , as it vanishes identically when the grains are not in contact . in this paper , we report results and analyses of a toy model for the compaction of a granular material .
our model consists of `` grains '' placed on a square lattice and connected by vertical springs .
the grains are constrained to move only in the vertical direction , so the model treats forces and stresses as _ scalar _ quantities .
( see fig .
[ fig : model ] . ) geometric disorder is introduced in the model by assigning to each spring an equilibrium length drawn randomly from a square distribution .
though we can not hope to capture the true physics of tensorial stresses in this manner , we can study the validity of arguments that have been applied to real granular systems and should equally well apply to our toy model .
in particular , we compare the distribution of inter - grain forces generated by our model to the distribution predicted by the @xmath0-model of coppersmith et al @xcite , and we compare the stress - strain power law generated by our model to that expected on the basis of a mean - field argument .
our results provide a warning : in neither case are the theoretical predictions borne out by the numerics .
the structure of our model is shown in fig . [ fig : model ] .
grains are placed at the even - parity sites of a @xmath1 cubic lattice with lattice constant @xmath2 .
we refer to such a system as an `` @xmath3 '' system . excluding the sites on the lower boundary
, it contains @xmath4 grains .
grains are constrained to move only in the vertical direction , as if sliding on fixed , frictionless wires ( shown as thin lines in the figure ) .
we let @xmath5 denote the vertical displacement of the grain at lattice site @xmath6 from its nominal height @xmath7 , with @xmath8 for upward displacements . between every two nearest - neighbor grains
there is a vertical spring .
all of the springs respond linearly under compression , with spring constant @xmath9 , but they produce no force under extension .
the energy of the spring between grains @xmath10 and @xmath11 is taken to be @xmath12 where @xmath13 and @xmath14 are independent , quenched random variables designating the different uncompressed lengths of the springs .
we take @xmath13 and @xmath14 to be distributed uniformly over the interval @xmath15 .
note that when @xmath16 , the uncompressed spring has length @xmath2 , the lattice constant .
one may imagine the grains to have arbitrary shapes and the uncompressed spring length to be the vertical spacing between grain centers when the grains are just barely in contact .
the system is taken to satisfy periodic boundary conditions in the horizontal direction .
all grains at the top are constrained from above by a rigid `` ceiling '' and therefore satisfy @xmath17 . on the bottom row , grains are constrained by a rigid `` floor '' whose height @xmath18 is an independent variable ; thus @xmath19 .
let @xmath20 denote the maximum value of @xmath18 such that no springs are compressed .
typically @xmath20 , which is a function of the random @xmath21 s , is negative and of the order of @xmath22 in absolute value .
( it can not be less than @xmath23 . ) below we measure the floor height in terms of the incremental variable @xmath24 .
though we use the terms `` floor '' , `` vertical '' , etc . , for convenience , there is no gravity in the model and there is complete statistical symmetry between the up and down directions .
the algorithm discussed below for determining equilibrium configurations explicitly breaks this symmetry , though the configurations it produces do not .
our model apparently depends on the parameters @xmath25 and @xmath9 , but these merely set the scales for distance and force . for the rest of this paper , numerical values for distances
spring - length variations @xmath21 , grain displacements @xmath5 , and boundary displacement @xmath26 will be quoted in units of @xmath25 ; and forces individual forces @xmath27 and the total force @xmath28 on the ceiling in units of @xmath29 .
although the _ width _ of the distribution of spring lengths scales out of the problem , its _ shape _ could conceivably make a difference .
nevertheless , we do not expect the precise shape to be important as long as the probability of having very large uncompressed lengths decays sufficiently rapidly . in any event , we consider only the square distribution .
let us contrast the alternative of random spring constants with the random equilibrium lengths we study here .
if the @xmath9 s were random but all equilibrium lengths were equal , then as @xmath26 increased all the springs would begin to be compressed simultaneously at @xmath30 and the system would be perfectly linear for all @xmath31 ; the force - vs .- displacement law would be @xmath32 for @xmath31 and @xmath33 for @xmath34 , where @xmath35 is a constant determined by the random @xmath9 s . by contrast , with random spring lengths we capture some of the geometric disorder of real granular materials , which inevitably introduces nonlinearities associated with the formation of new contacts during compression .
we will refer to a spring under compression as an `` active bond '' , and to a site connected to such a spring as an `` active site . ''
an equilibrium configuration for @xmath26 positive , but not too large , has a mixture of active and inactive sites and bonds . for a given set of @xmath21 s ,
the equilibrium configuration of active sites and bonds is unique , although the exact positions of inactive sites are not determined .
this uniqueness follows from the strict convexity of the total energy as a function of the lengths of active bonds .
[ fig : configs ] shows three equilibrium configurations with the same set of @xmath21 s on a @xmath36 lattice , for different amounts of compression .
only the active bonds are shown , with line thickness proportional to the force transmitted .
also shown are two bars representing the ceiling and the floor .
let @xmath37 be the total force exerted by the system on the ceiling .
[ fig : onecurve ] shows a typical force curve for a @xmath36 system , together with the fractions of sites and bonds that are activated .
elementary considerations reveal the limiting behaviors of @xmath28 for large and small @xmath26 . for very small
@xmath26 , the force on the ceiling is due to a single chain of activated springs that reaches from top to bottom .
this chain , which follows the directed path @xmath38 through the lattice that yields the highest value of @xmath39 , produces a force linear in @xmath26 : @xmath40 when @xmath26 is very large , all of the springs in the system are compressed .
the system is then again entirely linear : because all of the springs have the same stiffness , @xmath41 where @xmath42 is a ( negative ) constant that depends on the random @xmath21 s and is difficult to compute . in general , as @xmath26
is increased from zero , additional chains are activated and @xmath37 increases .
in fact it can be shown that @xmath28 must increase monotonically with @xmath26 .
( this is why @xmath42 must be negative . )
surprisingly , however , the slope of @xmath37 may _ decrease _ with increasing @xmath26 .
occasionally bonds that are active at small values of @xmath26 become inactive for larger @xmath26 , thus reducing the effective spring constant of the network .
the simplest example of this behavior is illustrated in fig .
[ fig : break ] .
the figure shows a sequence of configurations in which the central bond is originally active , but becomes inactive due to the effective stiffening of other parts of the network when additional bonds are activated .
the algorithm we employ for generating equilibrium configurations of our model relies heavily on the fact that for a given set of active bonds , the system response is linear . by solving for the times at which inactive bonds are activated during compression
, the algorithm generates the entire curve @xmath37 , starting from @xmath43 and explicitly visiting every configuration . to construct the initial , uncompressed configuration ,
we begin with @xmath44 for all @xmath45 and march up layer by layer applying the rule @xmath46 this configuration can be thought of as the packing that would result from an infinitesimal gravitational force acting on the system .
( strictly speaking , we set @xmath47 , march upward , determine @xmath20 from the height reached , and translate all displacements accordingly . ) a typical initial configuration is shown in fig .
[ fig : trees ] .
sites joined by line segments are separated precisely by the equilibrium length of the spring between them .
springs that are not long enough to connect their two sites are not drawn .
already one can see that there is a rich geometric structure hidden in this model , quite similar to that described by roux et al @xcite .
the heavy lines in the figure highlight structures we call `` trees '' , consisting of all sites that are connected to the same site on the first layer .
we refer to the distance between the upper end of an _ inactive _ spring and the site to which it will eventually connect as a `` gap '' .
note that sizes of the gaps reflect the structure of our initialization algorithm .
we have moved all the grains to their lowest possible position , concentrating all of the small gaps that might be present in a chain into a single large gap at the top of a branch of a tree . to generate the force curve ,
we maintain a list of active sites and their active bonds , and we keep track of changes in the tree structure of the uncompressed chains as the floor is raised . for any fixed set of active bonds ,
the @xmath5 s increase linearly with @xmath26 , since all of the active springs are linear .
it is therefore a straightforward linear problem to solve for the rate , @xmath48 , at which each @xmath5 advances .
advancing @xmath26 through one stage involves solving for @xmath5 for some arbitrary increase in @xmath26 , for which we employ an iterative biconjugate gradient routine @xcite , using the current configuration of @xmath5 s as an initial guess for the solution .
the solution is used to determine the rates of advance of all the active sites .
the rate of advance of each inactive site is equal to that of the active site that supports it through an uncompressed chain .
the rates are used to determine the value of @xmath26 at which the network topology changes .
the @xmath5 s are then updated according to the calculated rates , the list of active sites or the tree structure is updated , and the process is repeated .
the updating of the network topology is necessitated either by the closure of a gap or by the breaking of a bond .
the closure of a gap can generate one of two types of events .
first , it can cause an additional set of sites to be activated , which increases the size of our active site list ( and also increases the size of the linear problem that must be solved on the next iteration ) .
second , it can result in a `` push '' event , in which an inactive branch of one tree is simply transferred to another tree without becoming active .
pushes dominate the behavior of the system in the early stages of compression , becoming more and more rare as compression continues and the number of inactive sites decreases .
the breaking of a bond , as mentioned in section [ subsec : features ] , is the final possibility for changing the network of active bonds .
breaks are generally rare events . during full compression of a @xmath36 system , which undergoes approximately 2000 events ( pushes , chain additions , breaks ) ,
there are typically about 5 breaks .
the algorithm reaches completion when all bonds have been activated . as explained above , further compression would be homogeneous with all @xmath27 s increasing at precisely the same rate . to compress one configuration on a 40x40 lattice , generating one complete force curve ,
requires approximately one hour of computation on a typical 200mhz workstation .
we have accumulated data for various lattices with at most a few thousand sites .
future efforts to optimize the algorithm should permit investigation of substantially larger lattices . despite our reliance on linear methods for evolving the system between gap events , it must be emphasized that the absence of tensile forces introduces a strong nonlinearity for larger increments in @xmath26 .
an alternative approach to following the entire evolution of the system would be to solve directly for the configuration of the system at an arbitrarily chosen value of @xmath26 by minimizing the nonlinear energy function of eq .
( [ eq : eij ] ) . such an approach might speed up the calculation even with linear springs , and it would be essential if the springs were nonlinear ( e.g. hertzian ) under compression .
fig [ fig : forcecurves ] displays numerically computed curves of @xmath28 vs. @xmath26 for a few rectangular systems of various sizes . at a cursory level , the force curves appear to have the form @xmath49 , which correspond to lines of slope @xmath50 on our log - log plots .
interestingly , this exponent depends strongly on system geometry . closer inspection of fig .
[ fig : forcecurves ] reveals both expected and unexpected behavior . for any particular realization of the disorder
, there is a region , sometimes substantial , on the log - log plot for small values of @xmath26 in which the force curve appears nearly linear because it is dominated by a single chain .
there is also a crossover to linear behavior for large @xmath26 since , as the fraction of active bonds approaches unity , the system approaches the linear limit described in previous sections .
the intermediate regime is the one of interest , and the behavior there is rather complex . in
many runs an exponent @xmath51 in the intermediate regime can be readily identified .
however , in other runs ( see , e.g. , fig .
[ fig : forcecurves]c ) , small - system statistics tend to obscure the phenomena .
it appears that completely reliable measurements of the exponents in the various regimes will require bigger systems , beyond the reach of our current numerical codes .
nevertheless , we believe that the results for systems of a few thousand grains support the conclusions drawn below .
( see ref @xcite for a discussion of a closely related model . ) in studying force curves for bead packs , several authors have proposed a mean - field argument suggesting that for various granular systems one should observe @xmath52 , where @xmath53 is the exponent of the single - contact force law .
thus for our model the mean - field theory predicts @xmath54 .
( for the reader s convenience , we summarize a version of the theory in the appendix .
the treatment there is similar to that of references @xcite and @xcite . )
the argument is based on the assumption that the rates at which pairs of nearest neighbor sites approach each other may all be taken to be equal to the average rate of compression . in subsection c below
we present numerical results for our model showing that @xmath50 can be significantly less than 2 for some system geometries and in subsection d we seek to explain the failure of the mean - field argument . in our view , this failure casts doubt on the argument s applicability to real granular systems .
useful supplementary information is covered in subsections b and e. it is instructive to consider two cases for which the force curves can be explicitly calculated . _ case 1 : _
this system consists of only a single layer of random - length springs , as shown in fig .
[ fig : limitmodels]a . assuming the equilibrium spring lengths to be uniformly distributed in a finite interval , one immediately obtains @xmath56 . in this case
the mean - field argument ( appendix a ) is exact , as all gaps do close at the same rate .
_ case 2 : _ @xmath57 . because of the periodic boundary conditions , this case is equivalent to a single column of grains without periodic boundary conditions , as shown in fig .
[ fig : limitmodels]b .
again we take the distribution of equilibrium lengths to be a square distribution of width @xmath58 . in this case
it is more convenient to fix the force @xmath28 and compute the displacement @xmath59 since the compressive stress on each grain must be the same .
let us decompose the displacement @xmath60 where @xmath61 is the displacement of the @xmath62th spring .
when @xmath63 , we have @xmath64 for each @xmath62 , the longer spring at each level being just at the threshold of compression .
the growth of @xmath61 with @xmath28 depends only on @xmath65 , the difference between the equilibrium lengths of the two springs in the @xmath62th layer .
specifically , @xmath66 note that the @xmath65 s are independent and all have the probability density @xmath67 for @xmath68 .
the expected value of @xmath69 is defined as @xmath70 using the fact that all the terms in the summation are equal and performing the integration , we obtain @xmath71 , which in turn may be integrated to yield @xmath72 where @xmath73 and @xmath74 are the nondimensionalized displacement and force per layer , respectively .
( [ eq : averagedelta ] ) shows that , in this case , there is no simple power - law behavior in the large @xmath75 limit .
note that eq .
( [ eq : averagedelta ] ) is approximately linear near @xmath76 ( i.e. , @xmath77 = 1 $ ] ) , and it is linear near and beyond @xmath78 .
if we attempt to identify a single power law for intermediate values of @xmath79 , the natural choice is the derivative @xmath80 , evaluated at the point where this quantity is most slowly varying ; i.e. , where @xmath81 vanishes .
this method yields @xmath82 with @xmath83 , where we have returned to displacement as the independent variable .
the analysis of case 2 shows that it is possible for the force curve to exhibit a region corresponding to a power less than 2 .
it also shows that the emergence of a true power law should not be taken for granted in these systems .
we find , however , that in sufficiently wide systems , a power law does arise ( cf .
discussion below ) .
_ short , wide systems : _ fig.[fig : forcecurves]a shows two typical force curves for an @xmath84 system .
it appears here that there is a regime in which @xmath85 , followed by the expected crossover to @xmath86 at large @xmath26 .
this observation lends some support to the mean field argument and is consistent with the claim of gilabert et al .
@xcite , who studied the electrical analogue of our model .
( see section [ sec : electrical ] below . ) _ tall , narrow systems : _ in a @xmath87 system , the exponent @xmath50 appears to shift noticeably .
the best fit to the curve shown in fig .
[ fig : forcecurves]b is @xmath88 in the intermediate regime of interest . the error in this measurement is estimated to be @xmath89 on the basis of fits made with different choices for which points to exclude from the intermediate regime .
the data clearly rule out @xmath54 .
_ roughly square system : _ an estimate of @xmath90 is obtained from data on @xmath36 systems .
[ fig : forcecurves]c shows data from two @xmath36 systems .
the apparent lack of consistent behavior is presumably due to finite size effects .
nevertheless , using data from 25 runs , an estimate of @xmath90 can be obtained as shown in fig .
[ fig : forcecurves]d . to obtain the dotted line in the figure ,
the data in the intermediate regime were fit to a power - law ; data from the initial linear regime and data from the two largest forces in the figure ( where we expect a crossover to linear growth ) were excluded in making this estimate for the exponent @xmath50 .
we suggest that the variation of exponent with aspect ratio can be traced to the tree structure in the system .
specifically , we conjecture that if most of the force is transmitted within a single tree , then a smaller exponent will be observed , whereas if the force is spread over many trees , a larger exponent .
we interpret the following numerical experiment as support for this conjecture . starting with a @xmath91 grid
, we removed all sites lying outside the @xmath92 cone emanating from the center of the bottom boundary .
we then measured @xmath37 for this reduced system . in this geometry all the active sites at every stage of compression are connected to the floor at the same point , so all the active bonds are contained within a single tree . as shown in fig .
[ fig : forcecurves]e the exponent @xmath50 is close to @xmath93 , indicating that the behavior within a single tree has a different character than in a system with many trees . to explore this issue
further , we measured the rate at which trees expand as a function of height _ in the initial configuration_. ( roux et al . have investigated a similar , but not identical , model @xcite . ) while building the initial configuration , we keep track of the root of the tree associated to each site , thereby counting the number @xmath94 of trees that survive up to layer @xmath95 . since only initial configurations are involved
not their subsequent evolution rather large systems can be simulated . for a system of width 50,000 with data collected for up to @xmath96 layers , we obtained an excellent power law fit @xmath97 , with @xmath98 and @xmath99 , with only a slight deviation for very small heights ( @xmath100 ) and very large heights ( corresponding to @xmath101 ) .
note that @xmath102 .
( incidentally , since @xmath103 , it is clear that the power law fit must be inaccurate for @xmath95 of order @xmath104 .
) generalizing from this system to one of arbitrary width , we estimate that in the initial configuration of an @xmath3 system , approximately @xmath105 trees will reach the top layer .
let us speculate on possible consequences of this estimate .
the rates of advance of all sites within a single tree are correlated , since compression of the bottom bond of the tree affects the rates of all the branches above it .
we therefore expect the sequence of gap closings to depend upon the initial tree structure .
assuming that the system obeys a simple anisotropic scaling law , we conjecture that in large systems the exponent @xmath50 is a function of @xmath106 alone .
limited support for this conjecture comes from two additional runs on @xmath107 and @xmath108 systems . in both cases , @xmath109 , and in both cases
we observe power - law behavior with @xmath110 , as shown in fig .
[ fig : forcecurves]f .
a more rigorous test would require runs on significantly larger systems , for which a more efficient code is needed .
note that the @xmath108 system was already too large for us to run to completion ; the curve in fig .
[ fig : forcecurves]f ends well before the crossover to the linear regime occurs .
note also that in the @xmath108 system , a clean power - law regime extends over at least two decades in @xmath26 and three decades in @xmath28 , giving us some confidence that a true power - law regime does exist in large systems .
a consequence of our scaling conjecture is that increasing the system size while keeping the aspect ratio _ fixed _ should result in exponents that eventually approach that of a wide , short system . as we have shown that @xmath54 in the @xmath111 limit ( case 1 , section ivb ) , we expect to observe the mean - field result @xmath54 in very large , approximately square systems as well .
the key assumption in the mean - field argument is that nearest neighbor sites approach each other at the mean compression rate . in terms of the probability density for inter - site distances ,
the assumption is expressed quantitatively in eq .
( [ eq : pxdelta ] ) of the appendix . which asserts that the shape and width of the distribution of inter - site distances are independent of the compression @xmath112 .
as discussed in the appendix , in our model the probability density @xmath113 is well defined only for @xmath114 ; i.e. , for bonds whose springs are compressed .
however , even restricting our attention to the active bonds , we find that our data are inconsistent with the above assumption .
specifically , in fig . [ fig : pofx]a we show the probability densities for the ( nonzero ) forces @xmath79 , where @xmath115 , at various stages of compression of a @xmath36 system , and it is clear that the distribution broadens as @xmath116 increases .
more quantitatively , in fig . [
fig : pofx]b we plot the widths of the best fits to the data by gaussian distributions restricted to @xmath117 , for several values of @xmath116 .
the broadening of @xmath113 , which occurs during the early and intermediate regime of compression , is a simple consequence of force balance at branching points of chains of active bonds .
consider a site at which three active bonds meet , forming a `` y '' ( either right - side - up or upside - down ) . because there are no tensile forces , the force in the unpaired branch of the y is greater than either of those in the paired branches .
moreover , as the y is further compressed , the rate at which the force in the unpaired branch increases must equal the _ sum _ of the rates of increase of the forces in the paired branches . since the larger force evidently increases more rapidly than the two smaller forces ( the forces almost always being increasing functions of @xmath26 )
, the distribution broadens . during the late regime of compression
, the addition of active bonds will convert y s to x s , for which any correlation between force and bond compression rate must be more subtle . indeed , when all of the bonds are activated , all bonds compress at exactly the same rate and the force distribution simply shifts uniformly to higher forces the mean field assumption becomes exact .
thus we expect to observe a broadening of the force distribution during early and intermediate stages of compression , with a rate that decreases as the density of y s becomes smaller .
this is precisely the behavior displayed in fig .
[ fig : pofx ] , where the width of the positive @xmath118 portion of the distribution is seen to grow roughly as a power of @xmath26 less than unity for small @xmath26 and level off at high compression . a quantitative calculation of the rate of broadening would require a detailed understanding of the statistics of branching in the active network and is beyond the scope of this work .
let us argue that the broadening of the force distribution generically leads to an exponent @xmath50 smaller than the mean field value .
suppose , for definiteness , that @xmath113 has a sharp leading edge at @xmath119 ; i.e. , that @xmath120 for @xmath121 and @xmath122 is bounded away from zero . the existence of y s in the active bond network leads to an advance of the leading edge @xmath119 that is more rapid than the advance of @xmath116 itself . taking into account the formation of new contacts and branch points during compression , and assuming simple asymptotic behavior
, we expect that @xmath123 with @xmath124 .
@xcite thus , to lowest order in @xmath116 , we find @xmath125 , and therefore @xmath126 .
corrections to this exponent should become significant for @xmath116 of order unity , which is also the order of the width of @xmath127 .
this expectation is consistent with fig .
[ fig : forcecurves ] , where it can be seen that the crossover to the linear regime for large @xmath26 occurs for @xmath128 , which is the last half decade in the plots .
these qualitative results do not depend on the linearity ( under compression ) of the springs in our model , nor do they depend on the dimensionality of the model .
they should also apply to the _ vertical _ forces in a system with horizontal degrees of freedom , providing only that the creation of new branch points is sufficiently common .
thus the mean - field argument seems problematic for physical bead packs .
the analysis shows that new contact formation can lead to exponents smaller than the mean field value , an important point in light of other mechanisms proposed to explain experimental observations of this exponent .
@xcite the analogy between the elastic properties of a network of linear springs and the electrical properties of the same network of resistors has been exploited in numerous studies @xcite . in order to clarify the relation between our model and others , and particularly with random networks near the directed percolation threshold ,
let us consider the analogy in some detail .
we will see that there are good reasons to be skeptical of the applicability of percolation model results to our model , but there is also an intriguing numerical coincidence .
there is a formal identity between our equations for mechanical equilibrium @xmath129 and the electrical equations @xmath130 in the latter equation , @xmath131 specifies the current between sites that are joined by a resistance @xmath132 in series with a battery generating a potential difference @xmath133 .
note that the resistors conductance plays the role of the springs stiffness .
the property that the springs function only under compression can be modeled in the electrical system by the insertion of perfect diodes , all directed `` downward '' , in series with each resistor .
roux et al . studied this very model @xcite , but did not report results for varying aspect ratios or single trees and did not study its relation to the @xmath0-model , which had not yet been introduced .
a system with batteries is substantially more complicated than a simple , randomly diluted resistor - diode network .
to relate the randomly diluted network at the directed percolation threshold to our model just beyond the initial linear regime , one must first assume that the network of active springs in our model has the same structure as the current carrying paths in a directed network of resistors placed at random on the lattice .
one must also assume a relation between @xmath26 of our model and the probability @xmath134 that a bond exists in the percolation system .
the natural assumption here would be that the probability @xmath135 is proportional to @xmath26 , where @xmath136 is the critical value for percolation , since @xmath43 is the point where a single force chain ( or current carrying path ) first forms and on average the bonds in our model are compressed by an amount proportional to @xmath26 @xcite . in our system , however , the batteries play two roles .
first , they generate potentials that affect the current distribution even when all the diodes are forward biased , which would correspond to a trivial , homogeneous state of the simple resistor network .
perhaps more importantly , however , they determine _ which _ diodes will be forward biased for a given applied potential difference across the whole network .
this dynamical process of selecting the current carrying paths may or may not yield structures well - modeled by the random addition of resistive links in a percolation model . in spite of the difficulties in establishing a connection
, it is interesting to compare our results for the power law obeyed by the stiffness @xmath137 to the conductivity exponent obtained from the theory of directed percolation in randomly diluted resistor networks @xcite . the conductivity exponent for directed percolation has been calculated both numerically and using renormalization group methods and appears to be approximately @xmath138 , @xcite which corresponds to a value of @xmath139 for the exponent @xmath50 .
the fact that this agrees with our measurements on tall , narrow systems deserves further study .
other authors have investigated additional details of the statistics near threshold in our model and argued that the system is closely related to the percolation one .
we now consider the statistics of forces transmitted by individual springs , exploring , in particular , the relation of our results to the @xmath0-model of coppersmith et al .
our model is simpler than the granular packings the @xmath0-model was intended to describe .
however , our model would appear to be a better candidate for description by the @xmath0-model than the original granular packings , since we have removed the tensorial stresses from our system but still have contacts whose formation is governed by quenched randomness . in the @xmath0-model in two dimensions
, @xmath140 refers to the fraction of the force on site @xmath6 from above that is transmitted to its neighbor at @xmath141 , the complementary fraction @xmath142 being transfered to site @xmath143 .
one assumes that each @xmath140 is a random variable that ( 1 ) is independent of the other @xmath0 s and ( 2 ) has the same distribution @xmath144 at every site , independent of the force supported by that site .
analytical studies of the @xmath0-model show that the distribution of forces supported by a single grain has an exponential tail at large forces whenever @xmath144 is non - vanishing at @xmath145 .
@xcite by contrast , the force distributions in our model , plotted in fig .
[ fig : pofx ] , show no evidence of exponential tails . at no stage during the compression does it appear that the @xmath0-modeldistributions are a good match for the distributions we observe . in particular , consider the fourth curve from the left in fig .
[ fig : pofx ] , which is made at a compression for which almost all _ sites _ are active but there remains an appreciable fraction of inactive _ bonds_. the distribution of @xmath140 s directly measured from our data in this regime ( see fig . [
fig : qdist]a ) is reasonably represented by @xmath146 + 0.8 - 0.2\cos(2\pi q),\ ] ] for which 20% of the @xmath0 s are either 0 or 1 .
the @xmath0-model would predict an exponential tail in the single grain force distribution .
@xcite even for the small systems we study here , the exponential tail would be clearly distinguishable from the rapid decay we observe . fig . [
fig : qdist]b compares numerical results from our model and from a simulation of the @xmath0-model with @xmath144 given by eq .
( [ eq : eta ] ) . for the @xmath0-model simulation , forces on the top row
were chosen randomly from a uniform distribution on the interval @xmath147 . to understand the discrepancy between our results and the predictions of the @xmath0-model
, we examine our data as regards assumption ( 2 ) above : i.e. , that the distribution of @xmath0 at each site is independent of the force supported by that site .
[ fig : qdist]c shows the distribution of @xmath0 s obtained from equilibrium configurations of our model corresponding to the same conditions as in fig .
[ fig : qdist]a , but separated according to the force supported by the site .
different symbols in the plot indicate different levels of force as described in the figure caption .
it is clear that the contribution to @xmath144 from larger forces is peaked more strongly about 1/2 and has little weight near @xmath148 or @xmath104 , which explains why the @xmath0-model prediction fails for this system . as a quantitative measure of the correlation ,
we have computed the covariance @xmath149 for our model , where @xmath150 and the averages are performed over space for one realization of a @xmath36 system .
( the square in this definition is necessary because left - right symmetry guarantees that a correlation function linear in @xmath151 would vanish . )
@xmath152 vanishes identically in the @xmath0-model since @xmath140 and @xmath153 are independent in that model . as shown in fig .
[ fig : qdist]d , @xmath152 develops a significant negative value when the system is under compression , indicating that larger forces are associated with @xmath0 s closer to 1/2 .
the heavy dot in the figure indicates the point corresponding to the data in parts ( b ) and ( c ) .
we briefly summarize the results of our study of the scalar model , discuss its generalization to three dimensions , and finally draw two conclusions concerning the implications for real systems or more realistic models . in the context of a toy model
, we have tested arguments that have been applied to stresses in static , non - cohesive granular materials .
further study of the model is needed , especially simulations of larger systems , but already two important facts have been established .
( 1 ) correlations in the stress configuration are responsible for substantial effects , both at the level of single - grain forces and that of macroscopic stresses .
( 2 ) depending on the ( properly scaled ) aspect ratio , new contact formation may play a decisive role in determining the macroscopic stress - strain relationship , with sufficiently tall systems showing power - law behavior with a nontrivial exponent . regarding ( 1 ) ,
we have identified two important effects : ( i ) the distribution of forces broadens under compression because of force balance constraints at sites where three active bonds meet ; and ( ii ) larger forces tend to divide more evenly between supporting grains as a result of the dynamical process that determines the structure of the active bond network .
theories that neglect these correlations fail when applied to our model
. one may wonder whether qualitatively new features might appear in a 3d generalization of our model in which the vertical direction is taken to be the 111 direction of a simple cubic lattice . as a preliminary check ,
we have measured the statistics of trees in the initial configuration and observed their general morphology .
the number of surviving trees decays as @xmath154 and the trees remain relatively compact . in other words ,
the diameter of surviving trees grows as @xmath155 ( compared to @xmath156 in two dimensions ) and the branches of separate trees do not become heavily entangled with each other . on this basis , we conjecture that the essential physics of the stress - strain relation will not be qualitatively different in three dimensions .
in particular , we expect a variation in the exponent @xmath50 with the scaled aspect ratio .
we draw two general conclusions .
first , it appears that the mechanism for the broadening of the vertical force distribution should be present in the full tensorial problem as well , and therefore should be re - examined as a possible explanation for the observation of smaller exponents than those derived from the mean - field argument .
experiments measuring the dependence of the exponent @xmath50 on aspect ratio would be especially interesting .
second , the success of the @xmath0-model in predicting the distribution of single - grain forces in bead packs stands in need of explanation .
the @xmath0-model has already been criticized for its neglect of tensorial stresses @xcite .
one may have expected it to provide a better description of the stresses in a scalar model of the type we have studied , but this is not borne out by our results .
we thank sue coppersmith and long nguyen for helpful conversations , and also scott zoldi for his suggestions .
this work was supported by nsf through grants dms-98 - 9803305 ( dgs and mgs ) and dmr-94 - 12416 ( jess ) .
10 h. m. jaeger , s. r. nagel , and r. p. behringer , physics today * 49 * , 32 ( 1996 ) .
s. n. coppersmith , c .- h .
liu , s. majumdar , o. narayan , and t. a. witten , physical review e * 53 * , 4673 ( 1996 ) , cond - mat/9511105 .
s. roux , d. stauffer , and h. herrmann , journal de physique * 48 * , 341 ( 1987 ) ; d. stauffer , h.j .
herrmann , and s roux , journal de physique * 48 * , 347 ( 1987 ) w. h. press , b. p. flannery , s. a. teukolsky and w. t. vetterling , numerical recipes in fortran ( cambridge , 1988 ) .
s. roux and h. herrmann , europhysics letters * 4 * , 1227 ( 1987 ) .
j. goddard , proceedings of the royal society of london , series a * 430 * , 105 ( 1990 ) .
a. gilabert , s. roux , and e. guyon , journal de physique * 48 * , 1609 ( 1987 ) .
if the active bond network were fixed , we would obtain @xmath157 , but the advance of the _ trailing _ edge of the distribution would invalidate the analysis ; for a fixed network of active bonds , the system is linear and @xmath86 .
p. de gennes , europhysics letters * 35 * , 145 ( 1996 ) . for a review , see ; e. guyon , s. roux , a. hansen , d. bideau , j .-
troadec , h. crapo .
reports on progress in physics * 53 * , 373 ( 1990 ) s. redner and p. mueller , physical review b * 26 * , 5293 ( 1982 ) .
b. arora , m. barma , d. dhar , and m. phani , journal of physics c * 16 * , 2913 ( 1983 ) .
s. roux , a. hansen , and e. guyon , journal de physique * 48 * , 2125 ( 1987 ) .
claudin , j .-
bouchaud , m. cates , and j. wittmer , physical review e * 57 * , 4441 ( 1998 ) , cond - mat/9710100 .
j. e. s. socolar physical review e * 57 * , 3204 ( 1998 ) , cond - mat/9710089 .
in this appendix , we summarize a mean - field argument that was developed to describe forces during the compression of a bead pack between parallel plates ( see , e.g. , refs . @xcite . )
we define the _ overlap _ between two adjacent beads as where @xmath158 is the distance between their centers and @xmath159 is the nominal distance at which the beads touch but exert no force .
note that because of the minus sign , the overlap is positive for beads in contact , and negative for beads not in contact .
we introduce the random variable , @xmath160 to be the overlap between two randomly chosen adjacent beads .
thus the sample space for @xmath160 includes both the choice of configuration ( spring lengths in our model ) , and the choice of a pair of adjacent beads .
of course @xmath160 also depends on the displacement @xmath26 of the floor plate .
( as in section i , we normalize so that nonzero forces start at @xmath43 ) .
we rescale the independent variable , defining @xmath161 where the bead pack is @xmath75 layers thick .
let @xmath113 be the probability distribution for @xmath160 .
the average force per spring , @xmath162 , is given by ; @xmath163 where @xmath164 . since @xmath165 the integral
may be restricted to @xmath166 .
of course , by our normalizations , @xmath167 so @xmath168 . the rescaled variable @xmath169 equals the average change in the overlap @xmath160 resulting from motion of the floor ; thus , @xmath170 where @xmath171 is the average overlap when @xmath172 .
the first main assumption of the mean - field argument is a considerable strengthening of ( [ eq : dmd0 ] ) : one assumes that the shape and width of the probability distribution are independent of @xmath116 , i.e. , @xmath173 substituting ( [ eq : pxdelta ] ) into ( [ eq : sigma ] ) and rescaling ( [ eq : fxo],[eq : pxo ] ) , we deduce that @xmath174 which is equivalent to eq .
( 1.2 ) of ref .
@xcite .
the second main assumption of the mean - field argument is that @xmath175 is continuous in both arguments and @xmath176 where @xmath42 is a finite positive constant .
combining the two assumptions , we find that @xmath177 dx,\ ] ] which immediately yields @xmath54 . under assumption ( [ eq : pxdelta ] ) , the only way to accommodate a value of @xmath50 less than 2 is to posit that the limit in eq .
( [ eq : pofo ] ) is infinite .
note that in our model , the distribution for @xmath160 is not uniquely defined for @xmath178 since inactive grains are free to relocate slightly .
despite this ambiguity , in section ivd we are able to test hypothesis ( [ eq : pxdelta ] ) by focusing on positive values of @xmath160 . | we study a scalar lattice model for inter - grain forces in static , non - cohesive , granular materials , obtaining two primary results .
( i ) the applied stress as a function of overall strain shows a power law dependence with a nontrivial exponent , which moreover varies with system geometry .
( ii ) probability distributions for forces on individual grains appear gaussian at all stages of compression , showing no evidence of exponential tails . with regard to both results , we identify correlations responsible for deviations from previously suggested theories . |
my viewpoint comes from a bottom - up approach @xcite to modeling quasicrystal structure and explaining their thermodynamic stabilisation .
that is , we start with ab - initio or pair potential based evaluation of the total energy , to capture the t=0 behaviour ; or perhaps md ( molecular dynamics ) and mc ( monte carlo ) simulation for @xmath0 . we inductively identify motifs and restrict our model , so it has freedom to explore only the states we know are comparatively good .
[ at this level , the model may be formulated in terms of tilings or clusters , but the discrete geometry is just standing in as a way to label the distinct low - energy atomic configurations . ]
new simulations are constructed , which can handle larger length scales because there are fewer degrees of freedom . at the moment , this appears to be the most direct approach to acertain whether energetic stabilisation ( e.g. an implementation of penrose s matching rules ) is ever relevant to real quasicrystals . in a metal
, realistic pair potentials have friedel oscillations [ fig .
[ fig : potcluster](a ) ] .
i will review why this is the real - space analog of the ` hume - rothery ' ( and related ) mechanisms in reciprocal space .
the energy from second - order perturbation theory , computed in reciprocal space , is @xmath1 where @xmath2 is the free electron dispersion , and @xmath3 is the [ short ranged ] potential for an atom to scatter an electron . [
the factor of 2 is from electron spin ; i do not encumber this schematic formula with multiple atomic species .
] so @xmath4 is most negative when @xmath5 , the atom density at a reciprocal lattice vector @xmath6 , is strong for @xmath7 ( near to spanning the fermi sphere ) : that was hume - rothery s criterion . if we now do the sum over @xmath8 in ( [ eq : e2 ] ) , lumping the prefactors and calling it ` @xmath9 ' , we get @xmath10 where @xmath11 are the atoms positions ; the second equality just comes from fourier transforming , [ recall @xmath12 . ]
the effective interatomic potential @xmath13 turns out to be the electron susceptibility @xmath14 , convolved twice with @xmath15 in three dimensions . now
the sharp fermi surface in reciprocal space induces friedel oscillations in real space a factor @xmath16 in @xmath14 which are inherited by @xmath13 .
thus , hume - rothery stabilisation is practically equivalent to saying the second minimum in @xmath13 is essential for determining the structure .
at least , friedel oscillations are important in the al - tm family [ i(alpdmn ) , d(alnico ) ] and in the frank - kasper family [ i(znmgre ) ] . in this conference ,
mihalkovi and widom @xcite assert that the embedded - atom potentials [ which lack friedel oscillations , but implicitly include multi - atom interactions ] work excellently for the i(cdca ) family . as a result
, the sites within a ` cluster ' are not governed by interactions with the other atoms in the cluster , but are a resultant of overlapping spheres representing the potential wells of second and third neighbour atoms , including those outside the cluster .
this seemed to me the only explanation for the pseudo mackay cluster in e.g. @xmath17-alcufe .
[ the outer shells have icosahedral symmetry , but the innermost one has a roughly 1/3 filling of 20 sites on 3-fold axes , where one geometrically would have expected an icosahedron ] .
the al atoms have rather weak interactions , and sometimes seem to behave almost like a fluid moving around a framework of fixed and well - ordered tm atoms
. simulations by cockayne and widom @xcite , henley _ et al _
@xcite , and hocker and ghler @xcite found frequent hoppings of the al atoms , so that one wonders whether to speak of well - defined sites .
the configurational entropy from this is obviously huge .
simulational approaches have scarcely scratched the surface as far as addressing thermally excited disorder , but i would say we know what to do .
the reason it has not been tried is that a thorough ab - initio - based study of a good quasicrystal has been carried out just a handful of times so far , and it is much simpler to do it at @xmath18 .
the simulation experiences have led me to believe that in the known quasicrystals , clusters do not have physical reality ( in the sense of having significantly stronger intra - cluster bonding ) .
yet clusters ( or tiles , which i consider to be a closely related concept ) are inescapable as a framework to organise our understanding of a structure .
a lesson we learned from more fundamental fields of 20th - century physics is that we all need frameworks to organise our thinking , but we forget that it is a coordinate system we impose to decribe phenomena , not a physical reality .
we should not be seduced by our tools and ascribe fundamental significance to these mental constructs .
let us instead be open to ` complementarity ' : that is , dissimilar descriptions may secretly be equivalent .
our experiences with decagonal structure models @xcite offer abundant examples in which the _ same _ structure may be expressed in terms of cluster @xmath19 on network @xmath20 , or of cluster @xmath21 on network @xmath22 . perhaps cluster @xmath21 is related to @xmath19 by some sort of inflation ; or perhaps @xmath21 is associated with the voids in the packing of @xmath19 and vice versa .
( an example is the complementarity of bergman clusters and mackay clusters in icosahedral quasicrystals . ) in a great many models , the cluster is a mathematical corollary of a set of inequalities which express the energy minimisation .
this appears most baldly , perhaps , in abstract tiling models where one demands to maximise the frequency of occurrence of cluster @xmath23 @xcite one presumably would say that cluster @xmath23 does have a physical content in this model .
but in many cases , this optimisation forces the presence of a much larger supertiling , and much larger clusters @xmath19 appear at its vertices .
thus , the appearance of the cluster is dependent on things that happen far away : a slight change in e.g. stoichiometry , and the clusters which dominated the whole image may dissolve in favour of some other motif , as a consequence of delicately competing energies .
a second way in which the cluster configuration depends on faraway atoms can be seen in al - tm quasicrystals , which are well described using pair potentials that have a strong second potential well due to friedel oscillations .
the fact that an atom occurs in a particular site within a cluster is then not mainly due to interactions with its neighbours in the cluster , but is the resultant of potentials from many more distant neighbours in the space surrounding the cluster , and which are sufficiently correlated that the potentials from them add constructively ( see fig .
[ fig : potcluster](b ) ) . in conclusion ,
my proposed operational definition of clusters is statistical : ` a pattern of atoms which is found in all examples of a given ensemble . '
an ensemble is implicit in anybody s definition of a cluster : to locate the boundary around a group of atoms , it must be possible to surround the cluster in more than one way ; one could not talk of , e.g. , the fcc lattice as being built from clusters .
the implicit ensemble might be either multiple occurrences of the cluster in a large unit cell , or various crystal phases that contain the same cluster , or a single simulation cell in which one enumerates all the low - energy structures .
one can not rule out a cluster description a priori as a way to map the energy landscape of a structure .
when i was more naive , in fact , i advanced a cluster model which assumed a certain cost of cutting the linkages between mackay clusters along 2-fold and 3-fold symmetry directions , so as to predict the equilibrium crystal shape @xcite .
although a certain knobbliness might be anticipated , still even if the crevices got filled by al atoms the energy cost might still be a linear function of the cluster - cluster linkages cut by the placement of the interface .
but in view of the experience mentioned in sec .
[ sec : cluster - deca ] , i no longer expect that cluster linkages govern the energy differences due to offsets or orientations of the interface .
nevertheless , one can imagine ` clusters ' emerging in the purely statistical sense expressed in the ` operational definition ' at the end of sec .
[ sec : cluster - deca ] . if a crack approaches similar groupings of atoms , presumably it tends to pass through them in similar fashions .
then the atoms which always find themselves together on the same side of the crack could be designated a ` cluster ' for the purposes of describing crack propagation .
notice that ( i ) such ` clusters ' need not be the same ones that are useful in describing the equilibrium ensemble and ( ii ) one would expect their shape to have less symmetry than the material itself , as it must depend on the orientations of the crack propagation and of the crack shear . if it happens that in fact the _ same _ clusters appear in cracks of various orientations , and for other physically defined ensembles , that would be justification to attribute a ` physical reality ' to the cluster but such tests seem possible mainly in simulations , not in experiments .
my motivation here is not in the physical consequences of phason elasticity , but in using elasticity as an indicator of the nature of the quasicrystal state . _
gradient - squared _ elastic free energies appear only in the _ random - tiling _ kind of phase , in the sense distinguished in sec .
[ sec : stabilisation ] , above .
matching - rule interactions would lead to an energy cost which is _ linear _ in the absolute value of phason strain components @xcite , presumably the same is true for any other interaction that has the same ground state ( or to one in the same ` local isomorphism ' class ) .
a small caveat should be offered .
the linear cost is related to the discrete hops that are mathematically unavoidable in structures with the usual quasicrystal space groups @xcite . on the other hand , for unusual non - symmorphic space groups ,
a ` continuous phason mode ' is possible @xcite which _ may _ exhibit gradient - squared elasticity in the ground state , in the same sense that one - dimensional incommensurate crystals may . but no plausible atomic model structure of this class has ever been exhibited , much less a set of interactions for which it is the ground state .
the experiments of de boissieu _ et al
_ @xcite on icosahedral phases are the only ones i know that support the validity of elastic theory and thus , implicitly , a random - tiling - like equilibrium state .
but i will not fully trust the elastic interpretation of diffuse scattering till the quantitative elastic constants agree ( at least in order of magnitude ) with a plausible simulation .
that has not happened yet : one reason is that the ` canonical cell ' tiling , which is the simplest way to make a well - specific ensemble for most cluster - based icosahedral models , @xcite is also the least tractable tiling to simulate @xcite ; a second reason is that it is nontrivial to extract an absolute scale of fluctuations from measurements of diffuse scattering . in the case of _ decagonals , _ no evidence of gradient - squared elasticity has ever been seen in experiments . the behaviour of the entropic elastic theory in decagonal random tiling models is not well understood , either .
thus , it seems more plausible to me that matching rules ( or the equivalent ) are realised in decagonal quasicrystals , than in icosahedral ones .
the two fundamental competing scenarios of the stabilisation of quasicrystals are not exactly ` entropy ' versus ` energy ' , as we often loosely say .
rather , the question is whether the model is in the qualitative class that contains quasiperiodic ideal tilings that have purely bragg peaks , or the class that contains the maximally random tiling in which long range order is an emergent phenomenon .
this distinction is a rigorous one from the viewpoint of statistical mechanics , because these two states are separated by a phase transition , but that is of no help in distinguishing them experimentally .
i have come to believe that ab initio modeling , though tedious on account of the many levels of description between microscopic and macroscopic @xcite , is the quickest path to a solid understanding of which scenario should apply to a particular given material .
it is sometimes objected that simulation is unfeasible for handling e.g. incommensurate modulations , when the effective repeat cell is far too large to simulate by brute force .
but there are generally ways to bridge to large scales by connecting the simulations first to a kind of continuum model .
( a valid analogy is that one can understand the geometry of a large soap bubble by evaluating its surface tension , which can be computed using a far smaller simulation cell . ) to get meaningful results , it is crucial ( and very difficult ) that the structure used in the modeling be made consistent with the hamiltonian assumed it must be the ground state ( in the @xmath18 case ) , or nearly so .
rather generally , al - tm quasicrystals seem to have a framework of well - fixed sites plus a scattering of sensitive sites . in @xmath17-almnsi ( and perhaps some others ) , these are the @xmath24 atoms of ref .
@xcite , located on 6d body centres in the hyperspace formulation , or centring bergman clusters in real space , which can alternate between al or tm atoms or vacancies . in @xmath25-alconi ( ni - rich or co - rich ) , there are al atoms in ` channels ' @xcite which undergo ( in simulated models ) intricate occupational / displacive orderings . at a higher level of description , such atomic orderings can be expressed as terms maybe the dominant terms in the effective tile - tile interaction ( whether or not it realises matching rules ) .
it was asserted that ` a considerable amount of chemical disorder is essential to a random - packing model .
perhaps this is based on a mistaken picture that random packings ( of clusters ) necessarily create atom conflicts that must be resolved in a context - dependent way .
but actually each _ tile _ in a tiling whether random or governed by local rules has a finite set of local environments , and the atom decoration is _ designed _ to fit well with every environment , without overlapping or conflicting atoms . in the simplest cases
e.g. the square - triangle tiling with dodecagonal symmetry , or the rhombohedral tiling with ico symmetry
there are two kinds of tiles , and the ( irrational ) ratio of their numbers is fixed by the symmetry . the decoration is deterministic , so there is a unique stoichiometry .
however , tilings related to real quasicrystals often have more than two tiles
e.g. the ` canonical cell tiling ' @xcite with ico symmetry , or the ` hexagon - boat - star ' tiling @xcite with decagonal symmetry . then within the sum rules fixed by symmetry
, one kind of tile can be traded for a combination of the others ; sometimes the decoration is such that this trade leaves the atom count unchanged @xcite , so our model is line compound in this case too . in the case that the atom content changed , though
, the alloy composition would indeed be variable ; but as elser once suggested , such a decoration is undesirable since phason dynamics would be coupled to , and slowed by , mass diffusion , contrary to the observed relaxation in good quasicrystals . note that although it has become attractive to reformulate matching rules in terms of a decoration that implements gummelt - like covering rules @xcite , i do not think this is likely to emerge from an atomistic model .
if a covering cluster emerges , it will be a sort of accident from the maximimation of some smaller non - overlapping cluster ( see my comments on clusters in sec .
[ sec : cluster - deca ] ) .
instead , i consider it much likelier that something emerges similar in spirit to penrose s arrow rules .
that is : the larger energy scales expressed by the fundamental atomic sites define a random - tiling ensemble ; then additional occupational orderings in relatively rare sites define ( something like ) a set of interacting arrows .
in fact , widom @xcite and coworkers have found ( in the past few years ) a matching rule _
almost _ equivalent to penrose s , in the decagonal @xmath25(alcuco ) , implemented in that case by an alternation alco / coal in the chemical occupancy of a pair .
the evidence is abundant that many quasicrystals are high - temperature phases , ergo stabilised by entropy but , in many cases , not the tiling configurational entropy , for that is too small .
so , the larger entropy of vibrations or of chemical disorder is the only candidate to affect the phase diagram . in the past , we brushed aside such entropy contributions , by claiming they must have a similar value in the quasicrystal phase as they do in the approximant phases .
the latter phases were assumed to be the immediate competitors of the quasicrystal phase in the phase diagram , so the difference in vibrational or substitutional free energies would largely cancel .
but perhaps we need to examine more carefully just how the quasicrystal ordering may affect these entropic terms .
i am grateful to marc de boissieu , veit elser , marek mihalkovi , walter steurer , and mike widom , for conversations and communications .
this work was supported by the u.s .
department of energy grant de - fg02 - 89er-45405 .
99 c. l. henley , m. mihalkovi , and m. widom , j. alloys compounds 342 : 221 ( 2002 ) .
m. widom and m. mihalkovi , `` structural model of @xmath17-cacd based on canonical cells and supported by total energy calculations '' , these proceedings ( phil . mag . | personal comments are made about the title subjects , including : the relation of friedel oscillations to hume - rothery stabilisation ; how calculations may resolve the random - tiling versus ideal pictures of quasicrystals ; and the role of entropies apart from tile - configurational . |
ever since the establishment of general relativity , the search for the true nature of gravitational field has never ceased . in einstein s geometric description , the gravitational force is regarded as an effect of curved spacetime .
nevertheless , einstein himself proposed in 1920 that `` according to the general theory of relativity space is endowed with physical qualities '' @xcite .
meanwhile , eddington suggested that the deflection of light in a curved spacetime can be imaged as a refraction effect in a flat spacetime @xcite . and
wilson related this refractive index to the variable specific inductive capacity of the `` ether '' near matter , i.e. , suggested an electromagnetic theory of gravitation @xcite .
later , in 1957 , dicke @xcite , pointed out that vacuum polarization @xcite can account for such `` ether '' .
recently , puthoff reemphasized such an interpretation of gravitational field , and suggested an exponential form of the vacuum dielectric constant or vacuum refractive index @xcite .
vlokh further investigated the constitutive coefficients that could characterize the space as matter , and proposed the concept of optical - frequency dielectric impermeability perturbed by a gravitational field @xcite .
the dielectric - like vacuum interpretation of gravitational field has led to successful analysis of the gravitational effects such as the gravitational red shift , the light deflection , the time delay , the gravitational lensing , and so on @xcite . in this paper ,
the relation between the gravitational field and the quantum vacuum will be discussed further .
first , the gravitational metric tensor will be analysed and a corresponding graded refractive index in the flat spacetime will be deduced .
this refractive index implies a variable dielectric tensor of the gravitational space , which is essentially related to the polarization of vacuum .
then , through the analysis of the intensity of vacuum polarization , the number density of the virtual charge pairs in vacuum will be found to be a function of the gravitational mass @xmath0 and the distance @xmath1 .
this analysis leads to a polarizable vacuum interpretation of the gravitational force .
finally , the anisotropic propagation of light in a vacuum will be discussed .
the general relativity states that spacetime will be curved by the gravitational matter .
a curved spacetime can be described by the four - dimensional riemann space , where the velocity of light in vacuum is defined as a constant @xmath2 .
the interval @xmath3 ( @xmath4 is the proper time ) between the two adjacent points in this space is @xmath5 where @xmath6 is the coordinates of the point , for example , @xmath7 ( @xmath8 denotes the coordinate time ) , @xmath9 , @xmath10 , @xmath11 , and @xmath12 is the metric tensor @xmath13 for a static and spherically symmetric gravitational matter with mass @xmath0 , the schwarzschild exterior solution gives the metric @xcite @xmath14 where @xmath15 , @xmath16 , and @xmath17 , differing from the radius @xmath1 in flat spacetime , is the radial coordinate of the metric , @xmath18 is the gravitational constant , @xmath19 and @xmath20 have the same meaning as those in the usual spherical coordinates .
the above equation gives the metric tensor @xmath21 in a rectangular coordinate system of flat spacetime , if we set @xmath22 ( where @xmath23 , differing to the constant velocity @xmath2 , is the velocity of light in vacuum at distance @xmath1 from the gravitational centre in flat spacetime ) , @xmath24 , @xmath25 , @xmath26 , then eq .
( 3 ) can be rewritten as @xmath27 and the metric tensor will be @xmath28 since the metric tensor of a flat spacetime satisfies @xmath29 , we have @xmath30 @xmath31 considering @xmath32 at infinity , eq .
( 8) gives @xmath33 or @xmath34 combining eqs .
( 7 ) and ( 9 ) , we get the refractive index @xmath35 or through eqs .
( 7 ) and ( 10 ) , we get @xmath36 the above results are in agreement with those obtained through an optical way @xcite .
if we set a new tensor @xmath37 , each of its elements is the square root of the corresponding element of @xmath38 , then we have @xmath39 in weak field approximation , @xmath40 , the tensor becomes @xmath41 the above result implies a variable dielectric tensor @xmath42 of vacuum around gravitational matter : @xmath43 where @xmath44 is the constant permittivity of vacuum without the influence of gravitational matter .
( 14 ) and ( 15 ) correspond to a graded refractive index of the vacuum @xmath45 and a variable light velocity in gravitational field @xmath46
it is noticed that the electric displacement vector * d * in a dielectric medium can be written as @xmath47 where * p * represents the polarization of dielectric medium , @xmath48 is the polarization of vacuum @xcite , * e * denotes the electric field intensity . for the vacuum around gravitational matter , the variable dielectric tensor expressed by eq .
( 15 ) indicates that the intensity of vacuum polarization will be @xmath49 as we know that , the intensity of dielectric polarization is @xmath50 where @xmath51 is the number density of electric dipoles in the dielectric medium , @xmath52 is the electronic charge , @xmath53 is the stiffness factor .
similarly , we suppose the intensity of vacuum polarization to be @xmath54 where @xmath55 is the number density of the virtual charge pairs @xmath56 in the vacuum at a distance @xmath1 from the gravitational mass @xmath0 .
thus we have @xmath57 or @xmath58 in general relativity , a clock at a distance @xmath1 from the gravitational centre is slower than that at infinity .
this effect of time dilatation can be understood as the frequency reduction of the virtual dipoles in gravitational field , that is @xmath59 where @xmath60 is the equivalent mass of a virtual dipole at distance @xmath1 , @xmath61 is the corresponding angular frequency at infinity . considering the energy ratio of a virtual dipole at distance @xmath1 to that at infinity @xmath62 we have @xmath63 where @xmath64 is the corresponding equivalent mass at infinity .
( 24 ) and ( 26 ) give the relation @xmath65 where @xmath66 is the stiffness factor at infinity .
therefore @xmath67 where @xmath68 is the number density of the virtual dipoles at infinity .
this result is very similar to the number density distribution of gas molecules in a gravitational field : @xmath69 where we have used the relations @xmath70 ( the potential energy of a gas molecule ) and @xmath71 ( the average translational kinetic energy of a molecule ) .
( 28 ) can be rewritten as @xmath72 where @xmath73 is the number of virtual dipoles in a volume of @xmath74 at infinity or in a volume of @xmath75 at a distance @xmath1 from the gravitational centre .
so we have the length relation @xmath76 which increases our understanding of the length contraction effect in gravitational field .
it is the effects of time dilatation described by eq .
( 24 ) and length contraction described by eq .
( 31 ) that leads to the slowing down of the light velocity @xmath23 in a gravitational field as expressed in eq .
( 17 ) , which , in general relativity , is regarded as a constant light velocity @xmath2 in a curved spacetime .
eq . ( 28 ) tells that the number density of the virtual dipoles in vacuum decreases with the distance @xmath1 from the gravitational centre .
this result indicates that there can be a polarizable vacuum interpretation of the gravitational force .
consider the number density of the virtual dipoles around a @xmath77-charged particle in a gravitational field.,title="fig:",width=259 ] consider the number density of the virtual dipoles around a @xmath77-charged particle in a gravitational field.,title="fig:",width=259 ] first we consider a @xmath77-charged particle located at a distance @xmath1 from a gravitational mass @xmath0 as shown in fig . 1 ( a ) . according to eq .
( 28 ) , the number density of the virtual dipoles around the charged particle is @xmath78\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \nonumber\\ = n_{\textnormal{p}'0 } \exp\left[\frac{3g}{c^2}\left(\frac{m}{\sqrt{x^2+(y+r)^2}}+\frac{m}{\sqrt{x^2+y^2}}\right)\right],\end{aligned}\ ] ] where @xmath79 is the mass of the particle , @xmath80 is the distance from the particle to a nearby point p(@xmath81,@xmath82 ) , @xmath83 is the distance from the point to the gravitational matter @xmath0 . the corresponding contour map of @xmath55 is shown in fig . 1 ( b ) , where a relatively central contour line has higher value of number density than an outer one . from the figure
we know that , for a given distance @xmath80 , the virtual dipoles in vacuum below the charged particle ( @xmath84 ) are denser than that above the charged particle ( @xmath85 ) . then there is a slightly higher probability for the charged particle to be coupled or `` attracted '' by the virtual dipoles at the lower side , which leads to the downward movement of the charged particle , or in other words , leads to the gravitational force on the particle . for an electrically neutral particle or neutral matter
, its inner distribution of positive and negative charges will in the same way lead to the effect of gravitational force .
once the gravitational metric tensor is associated with the dielectric tensor of the vacuum , it is natural to have the idea that the tensor may be in a more general form under certain conditions , that is @xmath86 if so , the vacuum in a specific gravitational field will exhibit its anisotropic characteristics in polarization .
for example , a vacuum in a nonstatic gravitational field may no longer keep the relation of @xmath80=@xmath87=@xmath2 and may behave somewhat like a birefringent crystal . in such a case , there may be an anisotropic propagation of light in the gravitational field .
the gravitational birefringence in a vacuum is an analogy of the photoelastic effect . in the latter case
, an isotropic material turns to be anisotropic and shows its birefringent characteristic in light propagation when a stress is applied to the material .
similarly , if altered by a specific electric or magnetic field , an isotropic vacuum may be turned into an anisotropic one just as an isotropic dielectric medium will be . in such a case
, there may be birefringence observed in the altered vacuum as the electro - optic or magneto - optic effect observed in a dielectric medium .
it is delightful that the anisotropic propagation of light in vacuum has attracted both theoretical and experimental interests in recent years @xcite .
it is hoped that substantial observational proof will be found for the birefringence of light in a gravity - or - electromagnetism - influenced vacuum .
the curved spacetime in general relativity is described by a gravitational metric tensor .
it is found that this tensor corresponds to a variable dielectric tensor of vacuum around gravitational matter if viewed in a flat spacetime .
in such a view , the vacuum possesses a graded refractive index , which slows down the velocity of light propagating to a gravitational body .
this medium - like property of gravitational space is attributed to the virtual charge pairs in vacuum .
the increasing number density of virtual dipoles towards to the gravitational centre naturally suggests a polarizable vacuum interpretation of the gravitational force .
in addition , a dielectric tensor of vacuum could be in a more general form than that with equal diagonal elements , which makes it possible that a vacuum under certain conditions turns to be anisotropic .
so , great importance is attached to the observations or experiments searching for anisotropic propagation of light in a vacuum modified by specific gravitational or electromagnetic fields . | the gravitational metric tensor implies a variable dielectric tensor of vacuum around gravitational matter . the curved spacetime in general relativity
is then associated with a polarizable vacuum .
it is found that the number density of the virtual dipoles in vacuum decreases with the distance from the gravitational centre .
this result offers a polarizable vacuum interpretation of the gravitational force .
also , the anisotropy of vacuum polarization is briefly discussed , which appeals for observational proof of anisotropic light propagation in a vacuum altered by gravitational or electromagnetic field . |
this work was supported in part by the doe , and case western reserve university .
i wish to thank cern for a stimulating environment while this work was completed , my collaborators on globular cluster age estimation , brian chaboyer , pierre demarque , and peter kernan , and steve weinberg for enlightening discussions on anthropic arguments in favor of a cosmological constant .
see e.g. , m.s .
turner , g. steigman , and l. krauss , _ phys .
lett . _ * 52 * , 2090 ( 1984 ) ; m.s .
turner , _ physica scripta _ * t36 * , 167 ( 1991 ) ; p.j.e .
peebles , _ astrophys .
j. _ * 284 * , 439 ( 1984 ) ; g. efstathiou et al .
, _ nature _ * 348 * , 705 ( 1990 ) ; l. a. kofman and a.a .
starobinskii , _ sov .
lett . _ * 11 * , 271 ( 1985 ) ; l. a. kofman , n.y.gnedin , and n.a .
bahcall , _ astrophys .
j. _ * 413 * , 1 ( 1993 ) .
l.m . krauss and m.s .
turner , _ gen .
* 27 * , 1137 ( 1995 ) j. p.
ostriker , p. steinhardt , _ nature _ * 377 * , 600 ( 1995 ) s. weinberg , _ rev .
phys . _ * 61 * , 1 ( 1989 ) .
b. chaboyer , p. demarque , p. j. kernan , l.m .
krauss , _ science _ * 271 * , 957 ( 1996 ) m. bolte , c. hogan , _ nature _ * 376 * 1995 m.w.feast , & r.w.catchpole , _ mon . not .
r. astron .
_ , * 286 * , l1 ( 1997 ) i.n .reid , astro - ph/9704078,_astrophys . j. _ , in press b. chaboyer , p. demarque , p. j. kernan , l.m .
krauss , astro - ph/9706128 , submitted to _ astrophys .
j. _ w. l. freedman , astro - ph/9612024 , to appear _
proceedings for `` critical dialogs in cosmology '' , , princeton 250th anniversary _ ed .
n. turok p.j .
kernan and l.m .
krauss , _ phys .
* 72*. 3309 ( 1994 ) .
a. songaila , l.l .
cowie , c. hogan , and m. rugers , _ nature _ * 368 * , 599 ( 1994 ) d. tytler , x.m . fan , and s. burles , _ nature _ * 381 * , 207 ( 1996 ) d. tytler , s. burles , and d. kirkman , astro - ph/9612121 l.m . krauss and p. j. kernan , _ phys .
b _ , * 347 * , 347 ( 1995 ) . c. copi , d.n . schramm , and m.s .
turner , _ science _ * 267 * , 192 ( 1995 ) .
et al _ , _ phys .
lett . _ * 75 * , 3977 ( 1995 ) s.d.m .
white et al . , _ nature _ * 366 * , 429 ( 1993 ) d.a .
white and a.c .
fabian , _ mon . not .
r. astron .
_ , * 273 * , 72 ( 1995 ) j.p .
henry , u.g .
briel , and p.e.j .
nulsen , _ astron .
astrophys . _ * 271 * , 413 ( 1993 ) a.e .
evrard , c. a. metzler , and j.f .
navaro , _ astrophys .
j. _ * 469 * , 494 ( 1996 ) ; a.e.evrard,_mon . not .
r. astron .
_ , in press , astro - ph/9701148 ( 1997 ) j.a . peacock and s.j .
dodds , _ mon . not .
soc . _ * 267 * , 1020 ( 1994 ) .
j.a . peacock , astro - ph/9601135 ( 1996 ) j.a . peacock and s.j .
dodds , astro - ph/9603031 ( 1996 ) a.r .
liddle , d.h.lyth , p.t.p.viana , m. white , _ mon . not .
soc . _ * 282 * , 281 ( 1996 ) a.r .
liddle , d.h.lyth , d. roberts , p.t.p.viana , _ mon . not .
soc . _ * 278 * , 644 ( 1996 ) p.t.p .
viana , a.r .
liddle , _ mon . not .
_ * 281 * , 323 ( 1996 ) a.r .
liddle , p.t.p .
viana , astro - ph/9610215 s.d.m .
white , g. efstathiou , c.s .
frenk , _ mon . not .
soc . _ * 262 * , 1023 ( 1993 ) m. white , e.f .
bunn , _ astrophys.j . _ , * 450 * , 477 ( 1995 ) n.a .
bahcall , x. fan , and r. cen , astro - ph/9706018 t. totani , y. yoshii , and k. sato , astro - ph/9705014 s. perlmutter _ et al _ , astro - ph/9602122 a.kosowsky , m. kamionkowski , g. jungman , d. spergel astro - ph/9605147 , _ nucl.phys.proc.suppl .
_ * 51b * , 49 ( 1996 ) l.m .
krauss , d. scott , j. silk , m. white , to appear .
martel , p.r .
shapiro , s. weinberg , astro - ph/9701099 ( 1997 ) | the lower limit on the age of the universe derived from globular cluster dating techniques , which previously strongly motivated a non - zero cosmological constant , has now been dramatically reduced , allowing consistency for a flat matter dominated universe with a hubble constant , @xmath0 . the case for an open universe versus a flat universe with non - zero cosmological
constant is reanalyzed in this context , incorporating not only the new age data , but also updates on baryon abundance constraints , and large scale structure arguments . for the first time , the allowed parameter space for the density of non - relativistic matter appears larger for an open universe than for a flat universe with cosmological constant , while a flat universe with zero cosmological constant remains strongly disfavored .
several other preliminary observations suggest a non - zero cosmological constant , but a definitive determination awaits refined measurements of @xmath1 , and small scale anisotropies of the cosmic microwave background .
i argue that fundamental theoretical arguments favor a non - zero cosmological constant over an open universe
. however , if either case is confirmed , the challenges posed for fundamental particle physics will be great .
6.5 in 8.5 in -0.25 in 2cm^-2 2cm^2 3cm^-3 3gcm^-3 # 1#1 # 1#23.6pt lawrence m. krauss@xmath2 + + ( submitted to _ science _ ) the cosmological model perhaps most strongly favored by the data over the past few years has involved a proposal which is heretical from an elementary particle physics perspective . in order to reconcile a flat universe favored by both inflationary models and by the longstanding flatness problem in cosmology with the apparent age of globular clusters , and the fact that many estimates for the clustered mass density on large scales suggest that insufficient non - relativistic matter exists to achieve a flat universe , the idea that the cosmological constant is non - zero has been invoked@xcite .
most recently the inclusion of additional arguments associated with the baryon density of the universe and large scale structure have further strengthened the case for a cosmological constant @xcite .
the problem with this from a fundamental perspective is that a cosmological constant-associated in modern parlance with a non - zero vacuum energy density in the universe on a scale which would be cosmologically relevant and yet still allowed today would take a value which is roughly 125 orders of magnitude smaller than the naive value one might expect based on considerations of quantum mechanics and gravity ( see for example @xcite ) .
this apparent discrepancy would involve the most extreme fine tuning problem known in physics , and for this reason many particle physicists would prefer any mechanism which would drive the cosmological constant to be exactly zero today .
the possibility that cosmology might force physicists to have to directly confront this longstanding issue in fundamental physics is exciting , but at the same time its potential significance warrants a careful and continued examination of the data which motivates the confrontation . in this regard , the recent reanalysis of globular cluster age estimates based in part on new parallax measurements obtained from the hipparcos satellite is very significant .
globular cluster ages are obtained by fitting the observed color - magnitude diagram to the predicted distribution for a system of stars of different masses which all form at the same time and which are then evolved on a computer to a certain age .
a number of different fitting techniques are employed , with different uncertainties . because theoretical models predict the instrinsic stellar luminosity , while measurements yield an apparent magnitude , in order to compare theory and prediction , the distance to the globular cluster
must be known . using an extensive monte carlo analysis approach begun several years ago@xcite it was demonstrated that it is precisely this distance determination which leads to the dominant uncertainty in the inferred age of the oldest globular clusters .
based on the distance estimators then available , a 95% lower limit of @xmath3 gyr was determined for the mean age of 17 of the oldest globular clusters . by comparison ,
the age of a flat - matter dominated universe with hubble constant @xmath4 km s@xmath5 mpc@xmath5 , is @xmath6 $ ] gyr , the difference between the lower and upper limits quantified the extent of the cosmic age problem " .
this analysis was based on normalizing the color - magnitude diagram by utilizing the inferred instrinsic magnitude of rr - lyrae stars at a mean metallicity of [ fe / h ] @xmath7 to fix the magnitude of the horizontal branch for the globular clusters under study . by comparing the difference between this magnitude , and the magnitude of the observed main - sequence turnoff point for these clusters to the predicted magnitude difference based on stellar evolution modelling , one infers the age of the clusters .
other analyses obtained similar limits ( i.e. see @xcite ) .
recently , following hipparcos parallax measurements both for cepheid variable stars and for various subdwarf main sequence stars ( i.e. @xcite ) , we have been prompted to reanalyze the various results on the distance scale to globular clusters , and the resulting age estimates @xcite .
the changes have been dramatic .
a systematic shift in the estimated distances to globular clusters has made it clear that the earlier apparent convergence of such estimates was fortuitous , and the distance data was , and still remains , dominated by systematic errors .
our new best fit age is slightly lower than our previous 95% lower limit , and the lower limit we obtained ( by accounting for the now - apparent systematic uncertainties in the rr lyrae distance estimators ) is @xmath8 gyr . even allowing for minimum time of @xmath9 gyr after the big bang for the galaxy to form ,
this new lower limit is consistent with a flat matter dominated universe hubble age if @xmath10 . while this might suggest that the need to consider a cosmological constant has now vanished , it is important to remember that the age problem was just one , albeit an important one , of several cosmological arguments which together pointed in this direction . thus one must combine this new age estimate with the other constraints , which themselves have evolved , in order to reanalyze this issue .
following @xcite , i display in figure 1a , for a flat universe with cosmological constant @xmath11 , the parameter space of @xmath12 , vs @xmath13 ( where @xmath14 is density and @xmath15 is the critical density for a flat universe , so that @xmath16 ) , showing the allowed region given new age constraint , @xmath17 gyr , along with the other new constraints i shall describe momentarily . for reasons which will also become clear shortly , i present in figure 1b , the allowed range in the same parameter space for an open universe .
the quoted upper limit of 15 gyr is obtained by taking the @xmath18 upper limit of @xmath19 gyr obtained from the analysis in @xcite and adding 1 gyr as an upper limit to the estimated time after the big bang before star formation began in what would become the halo of our galaxy .
based on recent observations of primeval galaxies at redshifts in excess of 3 , this seems a reasonable upper limit on the time before structures began to form .
the other constraints displayed in figure 1 come from a consideration of the two other independent fundamental sets of cosmological observables , the baryon content of the universe , and large scale structure .
i shall describe each in turn below .
first , however , it is worth pointing out that current estimates of the hubble constant have themselves evolved .
two years ago , there was apparent incompatibility between hst measurements based on cepheid distances to virgo , and those based on supernova type 1a distance measurements .
these two distance measures have been converging ( i.e. @xcite ) , so that now a range for @xmath20 of @xmath21 brackets both measurements .
the horizontal dashed lines in the figures display this presently preferred suggested range for @xmath20 . for over 20 years
a robust upper limit on the baryon density of the universe has come from considerations of big bang nucleosynthesis ( bbn ) .
the predicted primordial abundance of the light elements up to @xmath22 , all of which are known to be produced primarily during bbn and not in stars , is a function in each case of the baryon to photon ratio in the universe .
hence a detailed comparison of all the inferred primordial light element abundances with predicted abundances yields a restricted allowed range of baryon densities .
that is , of course if there is concordance . during the past two years
there has been a great deal of heat , and some light , shed in this area .
first , starting about two years ago , bbn predictions @xcite began to tighten the allowed range of @xmath23 .
it began to become clear that unless systematic uncertainties in @xmath22 were allowed for the upper limit on @xmath23 coming from @xmath22 alone was becoming dangerously close to the lower limit .
the most robust upper limit on @xmath23 , however has for over 20 years come from observations of deuterium .
deuterium is only destroyed in stellar processing , so any observation of the interstellar or solar system abundance would put a lower bound on the actual primordial abundance of deuterium . since the predicted deuterium bbn remnant abundance is a monotonically falling function of @xmath23 , a lower bound on deuterium translates into an upper bound on @xmath23 .
the observed instellar value of @xmath24 puts a limit @xmath25 .
this situation took on a new dimension two years ago with the first claimed observation of deuterium absorption lines in primordial hydrogen clouds illuminated by distant quasars @xcite .
this method in principle allows a direct determination of the primordial deuterium abundance , and hence a direct measure of @xmath23 .
the only problem is that shortly after the original measurement , which gave an anomalously high value for the primordial deuterium abundance , other observations gave a value almost one order of magnitude lower .
this value , @xmath26 @xcite , is almost equal to the lower limit quoted above , and suggests not only that @xmath23 is near its upper limit , but also that there has been little chemical evolution of @xmath27 .
moreover , it reinforces the requirement that their be systematic errors in the @xmath22 abundance estimates if there is to be concordance in bbn .
@xcite .
most recently , the original high deuterium observation has been withdrawn . however , at the present time , until more observations are made , it seems premature to require that @xmath23 is near its upper limit .
we have seen over and over again that systematic errors are the dominant contribution to our uncertainty in cosmological quantities , and thus large shifts within the allowed range are as likely as small shifts .
moreover , concordence between the deuterium estimates and the @xmath22 estimates requires some significant systematic error in one of these values .
nevertheless , it is reassuring that even with the new data , a robust estimate of allowed range of @xmath28 @xcite from bbn , allowing for maximal systematic uncertainties in all abundance estimates remains compatible with the data .
this range is slightly larger than another independent estimate @xcite but is more easily compatible with the new deuterium abundance observations .
even this relatively large upper bound on @xmath29 is clearly incompatible with a flat baryon dominated for any reasonable hubble constant .
what is more surprising , however , is that it seems incompatible with a flat universe at all unless the hubble constant is extremely small , or there remains a large unclustered component of the energy density of the universe .
this result arises from considerations of x - ray measurements of hot gas in rich galaxy clusters .
if the gas in these systems is in hydrostatic equilibrium , and the overall gravitational potential is relativity smooth , one can invert an x - ray temperature / luminosity profile to get both the mass in hot gas , which is by far the largest baryonic component to the mass of the cluster , and also the total mass of the cluster .
thus , one gets a direct estimate for @xmath30 , where @xmath31 represents the fraction of the closure density which is in the form of clustered mass in the universe . if one assumes , for example , that @xmath32 , and non - relativistic matter dominates , then @xmath33 . in 1993 ,
white _
et al _ @xcite argued that the estimate of @xmath34 obtained from the coma cluster was sufficiently large so that it would be incompatible with the bbn upper limit for a flat matter dominated universe unless the hubble constant were extremely small . since that time a number of analyses of a broader class of clusters confirms the large @xmath34 estimates
@xcite , and these were utilized in previous work to put constraints in @xmath35 and @xmath12 space @xcite .
recently , a comprehensive theoretical analysis of cluster modeling has been completed @xcite to explore the robustness of the x - ray interpretations .
there is remarkable consistency between the numerical profiles and the data for a value of @xmath36 .
a comparison of this value with the bbn constraint would put very severe limits on @xmath35 .
however , if one desires a more conservative upper limit on @xmath35 one should consider the lowest possible value of @xmath34 which would be consistent with the data .
this value assumes all baryons are in the form of hot gas in the cluster and is given by @xmath37 @xcite .
similarly , taking the largest value of @xmath34 consistent with all the data , and allowing for a baryon fraction as large as .013 from stars and dark baryonic halo objects yields an approximate upper limit @xmath38 for the range of @xmath12 values of interest here . combining these limits with the bbn bounds yields a conservative constraint @xmath39 .
this is displayed in fig 1 for both an open and flat universe with cosmological constant .
finally , the last set of constraints comes from considerations of large scale structure .
this is another area where progress has been made recently , in particular as a result of measurements of cmb anisotropies . for some time , various independent observations have suggested that while the clustered mass in the universe exceeds the upper limit on the baryon density coming from bbn , it nevertheless falls short of the closure density .
recent observations confirm this trend . nevertheless , virial estimates suggest that @xmath40 , and i adopt this conservative lower bound here , as displayed in fig 1 .
i shall utilize two additional large scale structure constraints in this analysis .
first , observations of galaxy correlations supply a constraint on the shape of the power spectrum of density fluctuations on large scales .
this is an extremely powerful constraint because it is related to the primordial power spectrum primarily by considerations of causality , and the density of clustered matter in the universe . while the result is somewhat model dependent , assuming a cold dark matter dominated universe and assuming a roughly scale invariant initial spectrum of density perturbations , as predicted by inflation and as allowed by cobe constraints on the cmb , peacock and dodds @xcite obtained the constraint @xmath41 , which has been used in previous analyses .
more recently both these authors @xcite have noted that non - linear effects might alter their results on short scales . removing the shortest scale points and refitting one obtains @xcite @xmath42 at the 95 per cent confidence level . here
n is the spectral index of the primordial density perturbations , which cmb observations suggest is between @xmath43 .
we display this constraint in both figure 1a and 1b , as it is insensitive to the presence or absence of a cosmological constant .
finally , recently a number of authors @xcite have examined the abundance of galaxy clusters , which constrains the magnitude of density fluctuations on intermediate scales . by comparing to the cobe normalized value on large scales @xcite
, one can put a constraint on @xmath35 vs @xmath12 which is complementary to the shape constraint described above , although it turns out to be provide limits which are quite similar . because the growth of fluctuations between cobe scales and galaxy cluster scales is dependent on the geometry of the universe , this constraint is slightly different for open vs flat cosmological constant dominated cosmologies .
we plot these constraints @xcite in figures 1a and 1b .
the interplay between all of these new constraints in the @xmath35 vs @xmath12 parameter space is quite significant .
most important , by shifting the allowed region in @xmath12 upward , the new age limit combines with the other constraints so that now a _ larger _ region of parameter space is allowed for an open universe than for a flat universe with cosmological constant . in particular , it is now clear that smaller values of a hubble constant yield a universe which is now _ too old _ in a cosmological constant dominated universe .
the significance of the shift in age estimate which has taken place can be seen in both figures , where a solid curve displays what used to the upper limit on the allowed phase space coming from the earlier age constraint . for an open universe
, this old limit was extremely constraining . in this regard ,
the recent decrease in estimates for the hubble constant coming from hst measurements is also relevant .
it is also clear that while globular cluster age estimates have relaxed the constraint on the overall matter density of the universe , the other updated cosmological constraints coming from large scale structure and baryon counting have solidified in a region in which a flat , matter dominated universe is marginally viable only for an extremely small hubble constant .
the debate thus appears to remain between an open universe and flat universe with cosmological constant .
on the basis of shear size of allowed parameter space , for the first time an open universe is more strongly favored .
there are , however , additional theoretical and observational data which bear on this conclusion .
first , several preliminary observations have been made which are claimed to directly constrain @xmath35 and @xmath11 which on balance may swing slightly in favor of a non - zero cosmological constant .
cluster evolution , in number density @xcite , and luminosity density@xcite , has recently been used to place constraints on both @xmath35 , and @xmath11 .
the former , also displayed in figures 1a and 1b , yields a bound on @xmath35 which is inconsistent with a flat universe , but reasonably independent of the presence or absence of a cosmological constant .
the latter however , is claimed to strongly favor a non - zero cosmological constant .
in fact , a _ lower bound _ on @xmath44 of 0.37 at the @xmath45 confidence level is claimed .
this result is displayed in figure 1a , but it should be taken as preliminary . to counterbalance it , recent observations of the luminosity - redshift relation for type 1a supernovae @xcite place a preliminary _ upper limit _ on the value of @xmath44 of 0.49 . while this result involves the first application of a new technique , the recent measurements of more high redshift supernovae should , within the next few years , provide the strongest constraints on a cosmological constant in the universe today .
eventual measurements of cosmic microwave anisotropies on small angular scales will allow a direct measurement of the cosmological constant , in addition to other fundamental cosmological parameters , at the 10 - 20 @xmath46 level ( i.e. @xcite ) .
a distinction can be made between the effects of a cosmological constant , and a non zero curvature . for a flat universe , increasing @xmath47 , which implies decreasing the ratio @xmath48 , causes the magnitude of the first doppler peak to increase , while its position is relatively unaffected .
an open universe , on the other hand , changes the position of the first doppler peak , because introducing a non - zero curvature alters the redshift - angular scale relation of the universe . in this
regard , present measurements of cmb anisotropies suggest a large doppler peak in the location expected for a flat universe .
thus , while extremely preliminary , cmb measurements on small angular scales may favor of a cosmological constant over an open universe @xcite . at present , it is clearly too early to choose one cosmological model over the other .
it is clearly getting increasingly difficult to find accord with a flat universe without a cosmological constant .
the question then becomes : which fundamental fine tuning problem is one more willing to worry about : the flatness problem , or the cosmological constant problem ? the latter involves a fine tuning of almost 125 orders of magnitude , if the cosmological constant is non - zero and comparable to the density of clustered matter today , while the former involves a fine tuning of perhaps only 60 orders of magnitude if one arbitarily fixes the energy density of the universe at the planck time to be slightly less than the closure density .
numerological arguments might thus suggest that one should be more prepared to give up flatness than a zero cosmological constant .
i claim however , that this argument is incomplete .
we have a perfectly good theory , involving physics well below the planck scale , for why we might live in a flat universe .
as long as there was an inflationary regime in the early universe , the universe generically is driven to be approximately flat to many more decimal places than are required to resolve the flatness problem . moreover ,
when considering possible particle physics models of the early universe , inflation seems to be ubiquitous .
the difficulty seems to be not how to get enough inflation , but rather how to end it . on the other hand , we have absolutely no theory of the cosmological constant at all .
other than vague _ a priori _ prejudice , there is no well defined physical argument at the present time suggesting a zero , rather than arbitrarily small value of this quantity .
moreover , the energy scale associated with a non - zero cosmological constant which dominates the universe today is not unusual .
it corresponds to the characteristic mass scale which is discussed for neutrino masses which might solve the solar neutrino problem .
also interesting , recent arguments suggest that if the laws of physics predict a distribution of universes , with randomly chosen values for the cosmological constant , then quantitative anthropic arguments make it not implausible that it should be observed to be comparable to the matter density in the universe today @xcite .
whatever one s views toward anthropic arguments , it is not clear that the same reasoning could be applied to the flatness problem .
precisely because we have physical laws which suggest the universe should be flat , i would argue the _ a priori _ probability distribution for the curvature parameter one might reasonably consider should be strongly peaked about zero , in which case anthropic arguments along the lines applied to the cosmological constant might not be as suggestive , to the extent such arguments are suggestive . in conclusion , we should know within a decade whether the cosmological constant is non - zero , and whether we live in a flat universe .
the recent resolution of the age problem has dramatically altered the case for a cosmological constant vs an open universe .
a combination of cosmological observations now allows a larger parameter space for an open universe than a flat universe with cosmological constant . while one might argue that theoretical prejudice still favors the latter , if either of these cases represents reality , the implications for fundamental particle physics will be profound . |
supersymmetry ( susy ) is believed to be the most promising candidate for physics beyond the standard model ( sm ) . among many attractive features
, supersymmetry can provide understanding for the mass hierarchy of the weak scale and the planck scale .
the existence of the susy particles ( sparticles ) at the weak scale can satisfactorily lead to the gauge coupling unification at a desirable scale .
the electroweak symmetry breaking is triggered radiatively by the large top - quark mass .
the weak - scale susy also necessarily leads to rich physics in the near - future collider experiments @xcite .
however , current supersymmetric models do not advance our knowledge of flavor physics beyond the sm , such as the fermion mass generation and quark / lepton mixings .
in fact , generic susy models often have arbitrary flavor mixings and mass parameters in the scalar quark ( squark ) and scalar lepton ( slepton ) sectors and one would have to assume certain symmetries to prevent large flavor - changing neutral currents ( fcnc ) @xcite . on the other hand ,
the flavor structure in the susy sector motivates us to seek for new physics , and any experimental observation on the fcnc processes beyond the sm would undoubtedly shed light on our understanding for flavor physics . the evidence of neutrino oscillations observed by the super - kamiokande experiments serves as a good example @xcite . besides the conventional studies for fcnc effects in susy grand unification models in processes @xmath4 @xcite , neutral meson oscillations @xcite , and @xmath5 @xcite at low energies
, it has been actively pursued to explore the possible signatures at future collider experiments for the lepton - flavor violating processes in susy theories .
due to the super - gim mechanism , it has been pointed out @xcite that for low - energy rare processes induced by sparticle loops , the rates are suppressed by @xmath6 ; while for the flavor changing signal from direct slepton production at high energy colliders the suppression factor is at most @xmath7 , where @xmath8 and @xmath9 are the sfermion mass and decay width .
it is found that there is great physics potential for the cern large hadron collider ( lhc ) and the next @xmath10 linear collider ( nlc ) to probe susy flavor physics through slepton oscillations @xcite .
in contrast , there are several advantages for considering an @xmath11 collider to study susy flavor physics .
first of all , lepton and electron - photon colliders often provide much cleaner experimental environment for new physics searches than hadron colliders .
second , in models with soft mass unification or with ( minimal ) gauge - mediation of susy breaking , the lightest neutralino ( @xmath12 ) is expected to be lighter than sleptons ( @xmath13 ) .
therefore , the sparticle production for @xmath14 would have lower mass threshold @xcite than those for @xmath15 .
third , even when slepton pair production is kinematically allowed there is a @xmath16-wave ( @xmath17 ) suppression of the cross section near threshold , and the event rates are correspondingly limited .
on the other hand , the cross sections for @xmath18 production are proportional to @xmath19 near threshold , so high production rates are achievable .
fourth , since there exists the net electron flavor in the initial state , the observation on a lepton flavor other than an electron in the final state could be a clear signature for the flavor oscillation , with less severe sm backgrounds . in this paper , we study the signals for slepton oscillation between the first generation and the other two generations at an @xmath20 collider and their sm backgrounds .
we consider an @xmath1 collider with 500 gev c. m. energy and an annual luminosity 50 fb@xmath21 @xcite , running at the @xmath20 mode @xcite . in section
ii , we first illustrate the general flavor mixing in the slepton sector for the minimal supersymmetric standard model ( mssm ) . we then present the slepton production mechanisms and cross sections with and without the flavor mixing . in section iii ,
we study the slepton oscillation signals and their sm backgrounds at an @xmath20 collider .
we find that for the gaugino - like neutralino and chargino , it is promising to probe the mass differences and mixing angles between the left - handed sleptons to good precision .
for the right - handed slepton sector , the results are comparable to the existing one at @xmath1 and @xmath2 linear colliders @xcite .
the sensitivity obtained here is better than the most stringent low - energy constraints for lepton - flavor violation from @xmath22 @xcite .
we summarize our results in sec .
the mass terms for the left - handed sleptons ( partners to the left - handed leptons ) , right - handed sleptons and left - handed sneutrinos can be generally written as @xmath23 where @xmath24 are the generation indices , and @xmath25 are mass matrices which can be diagonalized through the unitary matrices @xmath26 . in terms of the slepton
mass eigenstates , the lepton - slepton interactions with gauginos are flavor non - diagonal : @xmath27 which generally leads to lepton - flavor violating effects .
if the slepton mass differences are large and the lepton - flavor off - diagonal couplings ( @xmath28 ) are sizeable , eq . ( [ fcnc ] ) may directly lead to flavor - changing processes like @xmath29 where the flavor index @xmath30 , as depicted in fig .
1(a ) and ( b ) , in which the black dots denote the vertices with flavor changing interactions . on the other hand ,
if @xmath31 ( @xmath32 ) is nearly degenerate with other sleptons ( sneutrinos ) , as prefered by suppressing the fcnc and as constrained by @xmath22 data , then the processes ( [ process ] ) and ( [ process1 ] ) may mainly go through the flavor diagonal production @xmath33 followed by @xmath34 oscillation subsequently @xcite .
the two cross sections are then related by @xmath35 where @xmath36 is the transition probability of interaction eigenstates @xmath31 to @xmath37 via oscillation , and @xmath38 and @xmath39 the mixing angle between @xmath31 and @xmath37 .
the cross section formula for sneutrino oscillation is the same as eq .
( [ osci ] ) with the replacement of @xmath40 . for the flavor diagonal production of @xmath41 , the differential cross section summed over the chargino helicity
are @xcite @xmath42 the subscripts on @xmath43 and @xmath44 refer to the electron and photon helicities .
the angle @xmath45 specifies the chargino momentum relative to the incoming direction in the c.m .
frame , @xmath46 is the chargino velocity in the c.m .
frame , and @xmath47 .
the @xmath48 helicity amplitude is @xmath49-wave near threshold so the cross section of eq .
( [ nu1 ] ) is proportional to @xmath50 ; the @xmath51 helicity amplitude , which comes only from the @xmath52-channel diagram , is @xmath16-wave near threshold so the cross section of eq .
( [ nu2 ] ) goes like @xmath17 .
the scattering amplitude is proportional to the wino fractions @xmath53 of the matrix @xmath54 that diagonalizes the mass matrix ( the first index @xmath55 labels the chargino mass eigenstate @xmath56 and the second index @xmath57 refers to the primordial gaugino and higgsino basis @xmath58 , @xmath59 ) .
further , the @xmath60 state fixes the incoming electron chirality to be left - handed `` @xmath61 '' .
selectron - neutralino associated production @xmath62 proceeds via @xmath63-channel electron and @xmath52-channel selectron exchanges @xcite ; see fig .
1(b ) with @xmath64 . by eq . ( [ osci ] )
, we can get the cross section of @xmath65 .
the contributions from higgsino components ( @xmath66 ) of @xmath67 can be neglected and only the neutralino mixing elements @xmath68 and @xmath69 enter . after summing over the neutralino helicities , there are four independent helicity cross sections as the helicity of the @xmath70 ( @xmath71 ) matches that of the @xmath72 ( @xmath73 ) : @xmath74 the @xmath75 @xmath76 for @xmath71 @xmath77 are effective couplings given by @xmath78 , \quad f_{ir } = \frac{-z_{i1}^{*}}{\cos\theta_w}. \label{lrcouplings}\end{aligned}\ ] ] here the @xmath79 are the elements of matrices that diagonalize the neutralino mass matrix ( the first index @xmath55 labels the neutralino mass eigenstate @xmath80 , @xmath81 , and the second index @xmath82 refers to the primordial gaugino and higgsino basis ( @xmath83 ) ) .
the angle @xmath84 specifies the direction of the selectron with respect to the direction of the incoming electron in the c.m.frame , @xmath19 is the velocity of the selectron , and @xmath85 . due to the approximate decoupling of higgsinos from the electron ,
the cross sections for the processes ( [ process ] ) and ( [ process1 ] ) are only large when the charginos and neutralinos are mainly gaugino - like , namely @xmath86 and @xmath87 .
fortunately gaugino - like @xmath88 , @xmath89 and @xmath90 are highly favored theoretically for two reasons : ( _ i _ ) the radiative electroweak symmetry breaking in susy guts theories yields a large @xmath91 value if @xmath92 is bounded by the infrared fixed point solutions for the top quark yukawa coupling @xcite ; ( _ ii _ ) @xmath93 is strongly preferred for @xmath94 to be a viable cold dark matter candidate @xcite . in the rest of our paper
, we will thus concentrate on this scenario . for our illustrations
we choose the chargino / neutralino masses ( in gev ) @xmath95 and the mixing matrix elements @xmath96 for the neutralinos and @xmath97 for the chargino .
these parameters correspond to the following mssm parameters at the weak scale @xmath98 where @xmath92 is a little larger than that at the infrared fixed point value@xcite to get a heavier higgs mass to avoid the lep2 limit and the convention for sign@xmath99 follows ref .
@xcite . for slepton masses ,
we choose ( in gev ) @xmath100 our choices for the gaugino and slepton masses are consistent with renormalization group evolution @xcite to the electroweak scale , with the following universal msugra parameters @xmath101 from eqs .
( [ nu1][neuthel2 ] ) , we see the following features : due to the stronger diagonal couplings @xmath102 , the cross sections for @xmath103 ( mainly @xmath104 ) and @xmath105 ( mainly @xmath106 ) are significantly larger than the other selectron channels ; the weaker neutral current couplings and the more massive scalar propagator in selectron production make their cross sections smaller than those for sneutrino production .
we illustrate these features in four processes with larger cross sections in fig . 2 .
two non - zero @xmath107 helicity combinations are labeled in parentheses .
the solid curves are of @xmath49-wave near the threshold resulting in a higher cross section rate , and the dashed ones of @xmath16-wave .
this indicates the possible signal enrichment by properly choosing the helicity configuration for the beams .
the lower curves show the effects of convolution with the backscattered laser photon spectrum for the machine parameters as chosen in refs .
the effect is to decrease the cross sections by about a factor of two . at an nlc with c. m. energy 500 gev ,
the typical production sections are of order 1000 fb for @xmath108 , and about 100 fb for @xmath109 and @xmath105 , on which three channels we will focus for our phenomenology study in the next section .
.[br - bf ] sparticle decay modes and branching fractions for the representative parameter choice . here
@xmath110 generically denotes a quark and @xmath111 denotes a lepton ; fermion - antifermion pairs @xmath112 with net charge @xmath113 are denoted by @xmath114 [ cols="^,^",options="header " , ] generally speaking , due to the distinctive kinematical characteristics , the multiple - jet ( @xmath115 ) signals for @xmath116 and @xmath117 have favorable signal - to - background ratios .
the irreducible sm background processes are @xmath118 including the hadronic decay branching fractions of @xmath119 decays , we find that the cross sections of eq .
( [ back ] ) are less than 0.1 fb at @xmath120 500 gev .
we could therefore expect a very effective probe to the oscillation parameters for processes ( [ channel1 ] ) and ( [ channel2 ] ) .
however , the leading background to the @xmath121 signal of eq .
( [ channel3 ] ) is @xmath122 the total cross section for this process at @xmath120 500 gev is about two picobarns , and the oscillation signal would be swamped by it . on the other hand , this signal process involves a right - handed electron in the initial state while the background process involves only a left - handed one due to the @xmath123 exchange .
if high beam polarization @xmath124 can be implemented @xcite , then this sm background can be highly suppressed .
helicity argument shows that choosing the right - handed photon beam could also help further suppress the background survived from the impure @xmath125 reaction , without hurting the signal [ see fig .
we thus assume that the background eq .
( [ big ] ) can been reduced to a manageable level . in our subsequent numerical discussions , based on the signal branching ratios in table ii
, we estimate the signal efficiencies to be 40% , 15% and 80% for eqs .
( [ channel1 ] ) , ( [ channel2 ] ) and ( [ channel3 ] ) , respectively .
we consider an oscillation signal to be observable if the cross section after the inclusion of the efficiencies to be no less than 0.1 fb .
for instance , a 0.1 fb signal cross section and negligible backgrounds with a 50 fb@xmath21 luminosity would correspond to about 99% c.l . discovery . in figs . 3 , 4 and 5 , we illustrate contours of constant cross sections for signal processes in ( [ channel1 ] ) , ( [ channel2 ] ) and ( [ channel3 ] ) , respectively , in @xmath126 plane . when plotting each figure , we assume that just one mixing angle exits , _
i.e. _ , the contributions originating from other mixing angles are set to zero .
the ( thick ) solid curves in the figures present the constant cross section contours for @xmath127 and 1 fb .
the dashed lines correspond to the current bound on the branching fraction @xmath128 . a lower value of @xmath129 is also indicated by the ( thin ) solid line .
those curves are obtained with the susy parameters presented in the previous section . as showed in figs . 3 and 4 , achievable sensitivity in @xmath130 for the left - handed slepton sector
can be significantly better than the low - energy constraint from @xmath22 .
it is even more impressive to probe the oscillation parameters in the right - handed slepton sector if the beam polarizations can be implemented to suppress the backgrounds .
compared with the resultes at @xmath1 and @xmath2 colliders @xcite , the potential of probing mixing angle @xmath131 between @xmath132 and @xmath133 are comparable .
so far , we have only discussed the slepton oscillations between electron and muon flavors .
our analyses are essentially applicable for @xmath134 oscillation as well .
in fact , our calculations for the signal by eqs .
( [ osci])([neuthel2 ] ) and the backgrounds of eqs .
( [ back])-([big ] ) should be formally identical for the @xmath135 final state .
the only differences are : theoretically , the mass difference between @xmath136 and @xmath137 may be bigger than that of @xmath136 and @xmath138 due to the larger yukawa coupling running for @xmath137 .
experimentally , the @xmath135 lepton tagging is less efficient than that for @xmath139 .
this may lead to about a 80% reduction on the event rate .
nevertheless , we expect about same order of magnitude sensitivity for @xmath134 oscillation . due to the electron flavor in the initial state ,
the signal processes under discussion are not sensitive to the @xmath140 oscillation .
recent study @xcite showed that an @xmath1 collider could probe this oscillation to high precision .
we only considered a representative set of susy parameters . as long as the lighter neutralino and chargino are gaugino - like
as anticipated in msugra models and in the minimal gauge - mediated susy breaking model , our results should be generally valid .
we presented our study with the slepton masses approximately 270 gev .
this has been beyond the reach for an nlc with @xmath141 gev via @xmath15 . for a heavier slepton spectrum
, we would need a collider with a higher c. m. energy .
the current discussions on the linear collider parameters @xcite indicate the feasibility to extend the c. m. energy to 1.5 tev .
it is always most beneficial that the c. m. energy is just above the signal production threshold .
however , if the neutralino and chargino are higgsino - like , then their couplings to leptons are too weak to result in significant cross sections for the signal at an @xmath20 collider . in summary , we have demonstrated that for gaugino - like neutralino and chargino , the slepton oscillation effects may be observable at a 500 gev @xmath20 collider with a 50 fb@xmath21 luminosity .
the achievable sensitivity in @xmath130 for the left - handed slepton sector can be significantly better than the low - energy constraint from @xmath22 . in the right - handed slepton sector ,
our results are comparable to that for @xmath1 and @xmath2 colliders , if the beam polarizations ( @xmath142 and @xmath143 ) can be implemented to suppress the backgrounds
. it may even be possible for an @xmath20 collider to go beyond the probe at an @xmath1 collider because of the kinematics advantage .
our analyses are essentially applicable for @xmath134 oscillation , with a sensitivity of the same order of magnitude as that for @xmath144 .
we thank h .- c .
cheng and j. feng for discussions .
this work was supported in part by the u. s. department of energy under grant no .
de - fg02 - 95er40896 and by national natural science foundation of china .
further support was provided by the university of wisconsin research committee , with funds granted by the wisconsin alumni research foundation . for recent reviews on susy collider phenomenology , see _ e. g. _ , x. tata , uh-511 - 872 - 97 [ hep - ph/9706307 ] ; v. barger , madph-97 - 1011 [ hep - ph/9801440 ] ; j. f. gunion and h. e. haber , scipp-97 - 37 [ hep - ph/9806330 ] , to appear in _
perspectives on supersymmetry _ , g. l. kane , ed .
( world scientific , 1998 ) .
s. bertolini , f. borzumati , a. masiero and g. ridolfi , nucl . phys . *
b353 * , 591 ( 1991 ) ; j. wu , r. arnowitt and p. nath , phys . rev . *
d51 * , 1371 ( 1995 ) ; v. barger , m. berger , p. ohmann and r. phillips , phys . rev . *
d51 * , 2438 ( 1995 ) ; h. baer , m. brhlik , d. castano , x. tata , phys . rev . *
d58 * , 015007 ( 1998 ) .
f. borzumati and a. masiero , phys .
* 57 * , 44 ( 1986 ) ; f. gabbiani and a. masiero , nucl .
* b322 * , 235 ( 1989 ) ; r. barbieri , l. hall , a. strumia , nucl . phys .
* b445 * , 219 ( 1995 ) ; david w. sutter , hep - ph/9704390 .
f. cuypers , g.j .
van oldenberg , and r. rckl , nucl .
* b383 * , 45 ( 1992 ) ; h.a .
knig and k.a .
peterson , phys .
lett . , 110 ( 1992 ) ; d.l .
borden , d. bauer and d.o .
caldwell , slac preprint slac - pub-5715 ( 1992 ) ; t. kon and a. goto , phys .
b295 * , 324 ( 1992 ) ; d. choudhury and f. cuypers , nucl . phys . *
b451 * , 16 ( 1995 ) ; k. kiers , j.n .
ng and g.h .
wu , phys .
* b381 * , 177 ( 1996 ) .
ginzburg , g.l .
kotkin , v.g .
serbo and v.i .
telnov , nucl .
inst . and meth .
* 205 * , 47 ( 1983 ) ; h.f .
ginzburg , g.l .
kotkin , s.l .
panfil , v.g .
serbo and v.i .
telnov , nucl .
inst . and meth .
* 219 * , 5 ( 1984 ) . v. barger , m. s. berger and p. ohmann , phys
d 47 * , 1093 ( 1993 ) ; phys . rev .
* d49 * , 4908 ( 1994 ) ; v. barger , m. s. berger , p. ohmann and r. j. n. phillips , phys . lett .
* b314 * , 351(1993 ) ; p. langacker and n. polonsky , phys . rev . *
d50 * , 2199 ( 1994 ) ; w. a. bardeen , m. carena , s. pokorski and c. e. m. wagner , phys .
b320 * , 110(1994 ) ; b. schremp , phys . lett . * b344 * , 193(1995 ) . analyses of supergravity mass patterns include : g. ross and r. g. roberts , nucl . phys . * b377 * , 571 ( 1992 ) ; r. arnowitt and p. nath , phys .
* 69 * , 725 ( 1992 ) ; m. drees and m. m. nojiri , nucl .
* b369 * , 54 ( 1993 ) ; s. kelley , j. lopez , d. nanopoulos , h. pois and k. yuan , nucl .
b398 * , 3 ( 1993 ) ; m. olechowski and s. pokorski , nucl . phys . *
b404 * , 590(1993 ) ; v. barger , m. berger and p. ohmann , phys .
rev . * d49 * , 4908(1994 ) .
h. baer , f. paige , s. protpopescu and x. tata , in _ proceedings of the workshop on physics at current accelerators and supercolliders _ , eds . j. hewitt , a. white and d. zeppenfeld , argonne national laboratory ( 1993 ) . | we study the possible signature for lepton - flavor violation processes in the scalar - lepton sector of the minimal supersymmetric model at future electron - photon colliders .
we find that an @xmath0 collider can provide a good opportunity to probe the mass differences and flavor mixing angles between the sleptons of the first generation and the other two generations for the right - handed as well as the left - handed sector .
the sensitivity reach for lepton - flavor violation is comparable to that obtained at an @xmath1 and @xmath2 collider , and is significantly better than the current bound from low - energy rare processes such as @xmath3 . |
the variational monte carlo ( vmc ) technique is a powerful way to estimate the ground state of a quantum mechanical system .
the basic idea is that one can use the variational principle to minimize energy expectation value with respect to a set of parameters @xmath0 @xmath1 following the monte carlo method for evaluating integrals , one intreprets @xmath2 as a probability distribution function .
typically , one assumes a functional form for the trial wave function , @xmath3 and the numerical advantage is that one can evaluate the energy integral by simply evaluating @xmath3 .
the method becomes variational when one then adjusts the parameters to optimize the trial wave function .
since the spectrum of @xmath4 is bounded from below , the optimized trial wave - function provides a best approximation to the true ground state of the system .
however , since @xmath5 has a partner hamiltonian , @xmath6 ( @xmath7 being the kinetic energy operator ) in which the spectrum of @xmath8 and @xmath9 are identical for all states above the ground state of @xmath8 .
that is to say , the ground state of @xmath9 has the same energy as the first excited state of @xmath8 and so on .
this hierarchy of related hamiltonians and the algebra associated with the susy operators present a powerful formal approach to determine the energy spectra for a wide number of systems .
@xcite to date , little has been done exploiting susy as a way to develop new numerical techniques . in this paper
, we shall use the ideas of susy - qm to develop a monte carlo - like scheme for computing the tunneling splittings in a symmetric double - well potential .
while the model can be solved solved using other techniques , this provides a useful proof of principle for our approach .
we find that the the susy / vmc combination provides a useful and accurate way to obtain the tunneling splitting for this system and excited state wave function for this system .
while our current focus is on a one - dimensional system , we conclude by commenting upon how the technique can be extended to multi - particle systems and to higher dimension . in short ,
our results strongly suggest that this approach can be brought to bear on a more general class of problems involving multiple degrees of freedom .
surprisingly , the connection between the monte carlo technique and the susy hierarchy has not been exploited until this paper .
supersymmetric quantum mechanics ( susy - qm ) is obtained by factoring the schrdinger equation into the form @xcite @xmath10 using the operators @xmath11 since we can impose @xmath12 , we can immediately write that @xmath13 @xmath14 is the _ superpotential _ which is related to the physical potential by a riccati equation . @xmath15 the susy factorization of the schrdinger can always be applied in one - dimension . from this point on we label the original hamiltonian operator and its associated potential , states , and energies as @xmath8 , @xmath16 , @xmath17 and @xmath18 .
one can also define a partner hamiltonian , @xmath19 with a corresponding potential @xmath20 all of this seems rather circular until one recognizes that @xmath16 and its partner potential , @xmath21 , give rise to a common set of energy eigenvalues .
this principle result of susy can be seen by first considering an arbitrary stationary solution of @xmath8 , @xmath22 this implies that @xmath23 is an eigenstate of @xmath9 with energy @xmath18 since @xmath24 likewise , the schrdinger equation involving the partner potential @xmath25 implies that @xmath26 this ( along with @xmath27 ) allows one to conclude that the eigenenergies and eigenfunctions of @xmath8 and @xmath9 are related in the following way : @xmath28 @xmath29 for @xmath30 .
denotes the @xmath31th state associated with the @xmath32th partner hamiltonian with similar notion for related quantities such as energies and superpotentials . ]
thus , the _ ground state of @xmath9 has the same energy as the first excited state of @xmath8_. if this state @xmath33 is assumed to be node - less , then @xmath34 will have a single node .
we can repeat this analysis and show that @xmath9 is partnered with another hamiltonian , @xmath35 whose ground state is isoenergetic with the first excited state of @xmath9 and thus isoenergetic with the second excited state of the original @xmath8 .
this hierarchy of partners persists until all of the bound states of @xmath8 are exhausted .
having defined the basic terms of susy quantum mechanics , let us presume that one can determine an accurate approximation to the ground state density @xmath36 of hamiltonian @xmath8 .
one can then use this to determine the superpotential using the riccati transform @xmath37 and the partner potential @xmath38 certainly , our ability to compute the energy of the ground state of the partner potential @xmath21 depends on having first obtained an accurate estimate of the ground - state density associated with the original @xmath16 .
for this we turn to an adaptive variational monte carlo approach developed by maddox and bittner.@xcite here , we assume we can write the trial density as a sum over @xmath39 gaussian approximate functions @xmath40 parameterized by their amplitude , center , and width .
@xmath41 this trial density then is used to compute the energy @xmath42 = tr[\rho_{t}e(x ) ] = \langle v_{1}\rangle + \langle q[\rho_{t}]\rangle \end{aligned}\ ] ] where @xmath43 $ ] is the bohm quantum potential , @xmath44 = -\frac{\hbar^{2}}{2m}\frac{1}{\sqrt{\rho_{t}}}\nabla^{2}\sqrt{\rho_{t}}.\end{aligned}\ ] ] the energy average is computed by sampling @xmath45 over a set of trial points @xmath46 and then moving the trial points along the conjugate gradient of @xmath47(x).\end{aligned}\ ] ] after each conjugate gradient step , a new set of @xmath48 coefficients are determined according to an expectation maximization criteria such that the new trial density provides the best @xmath39-gaussian approximation to the actual probability distribution function sampled by the new set of trial points .
the procedure is repeated until @xmath49 . in doing so
, we simultaneously minimize the energy and optimize the trial function . since the ground state is assumed to be node - less
, we will not encounter the singularities and numerical instabilities associated with other bohmian equations of motion based approaches .
@xcite moreover , the approach has been extended to very high - dimensions and to finite temperature by derrickson and bittner in their studies of the structure and thermodynamics of rare gas clusters with up to 130 atoms . @xcite
as a non - trivial test case , consider the tunneling of a particle between two minima of a symmetric double potential well .
one can estimate the tunneling splitting using semi - classical techniques by assuming that the ground and excited states are given by the approximate form @xmath50 where @xmath51 is the lowest energy state in the right - hand well in the limit the wells are infinitely far apart . if we assume the localized states @xmath52 to be gaussian , then @xmath53 and we can write the superpotential as @xmath54 from this , one can easily determine both the original potential and the partner potential as @xmath55 while the @xmath16 potential has the characteristic double minima giving rise to a tunneling doublet , the susy partner potential @xmath21 has a central dimple which in the limit of @xmath56 becomes a @xmath57-function which produces an unpaired and node - less ground state . @xcite using eq .
[ chargeop ] , one obtains @xmath58 which now has a single node at @xmath59 . for a computational example
, we take the double well potential to be of the form @xmath60 with @xmath61 , @xmath62 , and @xmath63 which ( for @xmath64 ) gives rise to exactly two states at below the barrier separating the two minima with a tunneling splitting of 59.32 @xmath65 as computed using a discrete variable representation ( dvr ) approach.@xcite for the calculations reported here , we used @xmath66 sample points and @xmath67 gaussians and in the expansion of @xmath45 to converge the ground state .
this converged the ground state to @xmath68 in terms of the energy .
this , in itself is encouraging since the accuracy of typical monte carlo evaluation would be @xmath69 in terms of the energy .
plots of @xmath16 and the converged ground state is shown in fig .
[ fig1 ] .
the partner potential @xmath70 , can be constructed once we know the superpotential , @xmath14 . here
, we require an accurate evaluation of the ground state density and its first two log - derivatives .
the advantage of our computational scheme is that one can evaluate these analytically for a given set of coefficients . in fig .
[ fig1]a we show the partner potential derived from the ground - state density .
where as the original @xmath16 potential exhibits the double well structure with minima near @xmath71 , the @xmath21 partner potential has a pronounced dip about @xmath72 .
consequently , its ground - state should have a simple `` gaussian''-like form peaked about the origin .
once we determined an accurate representation of the partner potential , it is now a trivial matter to re - introduce the partner potential into the optimization routes .
the ground state converges easily and is shown in fig .
[ fig4]a along with its gaussians . after 1000 cg steps , the converged energy is within 0.1% of the exact tunneling splitting for this model system .
again , this is an order of magnitude better than the @xmath73 error associated with a simple monte carlo sampling .
furthermore , fig .
[ fig4]b shows @xmath74 computed using the converged @xmath75 density .
as anticipated , it shows the proper symmetry and nodal position . by symmetry ,
one expects the node to lie precisely at the origin .
however , since we have not imposed any symmetry restriction or bias on our numerical method , the position of the node provides a sensitive test of the convergence of the trial density for @xmath75 . in the example shown in fig.[fig3 ] , the location of the node oscillates about the origin and appears to converge exponentially with number of cg steps .
this is remarkably good considering that this is ultimately determined by the quality of the 3rd and 4th derivatives of @xmath76 since these appear when computing the conjugate gradient of @xmath21 .
we have tested this approach on a number of other one - dimensional bound - state problems with similar success .
having demonstrated that the susy approach can be used to compute excitation energies and wave functions starting from a monte carlo approach , the immediate next step is to extend this to arbitrarily higher dimensions . to move beyond one dimensional susy , ioffe and
coworkers have explored the use of higher - order charge operators @xcite , and kravchenko has explored the use of clifford algebras@xcite . unfortunately , this is difficult to do in general . the reason being that the riccati factorization of the one - dimensional schrdinger equation does not extend easily to higher dimensions .
one remedy is write the charge operators as vectors @xmath77 and with @xmath78 as the adjoint charge operator .
the original schrdinger operator is then constructed as an inner - product @xmath79 working through the vector product produces the schrdinger equation @xmath80 and a riccati equation of the form @xmath81 for a 2d harmonic oscillator , we would obtain a vector superpotential of the form @xmath82 let us look at the @xmath83 part closer : if we us the form that @xmath84 , then @xmath85 which for the 2d oscillator results in @xmath86 .
thus , @xmath87 which agrees with the original symmetric harmonic potential .
now , we write the scaled partner potential as @xmath88 this is equivalent to the original potential shifted by a constant amount .
@xmath89 the ground state in this potential would be have the same energy as the states of the original potential with quantum numbers @xmath90 .
consequently , even with the this nave factorization , one can in principle obtain excitation energies for higher dimensional systems , but there is no assurance that one can reproduce the entire spectrum of states .
the problem is neither hamiltonian @xmath9 nor its associated potential @xmath91 is given correctly by the form implied by eq .
[ 2dfactr ] and eq .
rather , the correct approach is to write the @xmath9 hamiltonian as a _
tensor _ by taking the outer product of the charges @xmath92 rather than as a scaler @xmath93 . at first
this seems unwieldy and unlikely to lead anywhere since the wave function solutions in the second sector are now vectors rather than scalers .
however , rather than adding an undue complexity to the problem , it actually simplifies matters considerably .
as we demonstrate in a forthcoming paper , this tensor factorization preserves the susy algebraic structure and produces excitation energies for any dimensional susy system .
moreover , this produces a scaler@xmath94 tensor @xmath94 scaler hierarchy as one moves to higher excitations.@xcite
in brief , we have used the ideas of susy quantum mechanics to obtain excitation energies and excited state wavefunctions within the context of a monte carlo scheme .
this was accomplished without pre - specifying the location of nodes or restriction to a specific symmetry . while it is clear that one could continue to determine the complete spectrum of @xmath8 , the real challenge is to extend this technique to higher dimensions .
furthermore , the extension to multi - fermion systems may be accomplished through the use of the gaussian monte carlo method in which any quantum state can be expressed as a real probability distribution .
@xcite we offer this paper as the starting point for stimulating interest in developing numerical techniques based upon susy quantum mechanics .
this work was supported in part by the national science foundation ( erb : che-0712981 ) and the robert a. welch foundation ( erb : e-1337 , djk : e-0608 ) .
the authors also acknowledge prof .
m. ioffe for comments regarding the extension to higher dimensions . | we present here a supersymmetric ( susy ) approach for determining excitation energies within the context of a quantum monte carlo scheme . by using the fact that susy quantum mechanics gives rises to a series of isospectral hamiltonians ,
we show that monte carlo ground - state calculations in the susy partners can be used to reconstruct accurately both the spectrum and states of an arbitrary schrdinger equation .
since the ground - state of each partner potential is node - less , we avoid any `` node''-problem typically associated with the monte carlo technique . while we provide an example of using this approach to determine the tunneling states in a double - well potential , the method is applicable to any 1d potential problem .
we conclude by discussing the extension to higher dimensions . |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.